Enhancing Graduate Education Assessment: A Machine Learning Approach to GPA Prediction for Medical Students

Enhancing Graduate Education Assessment: A Machine Learning Approach to GPA Prediction for Medical Students | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Enhancing Graduate Education Assessment: A Machine Learning Approach to GPA Prediction for Medical Students Wenyi Lai, Jinna Lin, Kailiang Shen, Zhihai Ling, Ying Guan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7186245/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 10 Feb, 2026 Read the published version in BMC Medical Education → Version 1 posted 12 You are reading this latest preprint version Abstract Background In recent years, Chinese medical postgraduate education has undergone significant transformation, with enrollment soaring to 156,000 students in 2023, accounting for 12% of the nation’s total postgraduate admissions. Recognizing high-achieving students at an early stage and learning from their success can change the way to influence future educational dilemmas. While the existing evaluation system remained limited in its ability to prospectively predict academic performance variability. This critical gap underscores the need for innovative, data-driven approaches to transcend conventional assessment paradigms. Our study used machine learning techniques to predict the academic outcomes (GPA) of medical postgraduate students in a certain university, providing evidence-based strategies that could be used to improve educational practices and improve student performance. Methods In this study, we worked with 1,133 postgraduate students at Southern Medical University (2020 cohort) while analyzing 42 variables, including demographic, undergraduate performance, postgraduate transition variables, and measures of their self-assessment. Using the Boruta algorithm, we identified the most important predicting features and then tested eight learning machine models to find the best one. Furthermore, we applied SHAP (Shapley Additive Explanations) to derive interpretable insights into the most critical features of success. Finally, we submitted this work as an interactive web application that allowed academic leaders to predict their students’ GPAs and provide much-needed proactive support. Results XGBoost model crushed the competitions, delivering higher predictions (AUC = 0.744, Accuracy = 82.8%, F1 = 0.902). SHAP analysis exposed the secret formula for success. A student’s career ambition, undergraduate ranking, and core subject mastery weren’t just important, but they were also game changers. Based on the results of our study, we built a smart web tool that could turn data into action, giving educators a crystal ball to flag high-potential students early, personalize support proactively, and allocate resources smartly. Our results were not just number crunching; it will also be a new playbook for medical postgraduate students’ education. Conclusion This research showed that XGBoost model did not just predict academic performance; it also revealed hidden pathways for students’ success. We found that students should focus on ambition, track record, and their mastery of core subjects to drive their success based on machine learning and SHAP models. More importantly, we developed a clever, user-friendly tool that might help educators identify students with high potential earlier and intervene with support ahead of challenges. GPA prediction Boruta algorithm SHAP model machine learning XGBoost model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background Recently Chinese medical postgraduate education has changed extraordinarily in recent years with enrolments reaching 156,000 students in 2023, representing 12% of postgraduate admissions nationally [ 1 ]. The current two-step evaluation system (standardized assessment combined with comprehensive re-assessment) is effective for assessing theoretical knowledge and clinical competencies, but lacks predictive capabilities in terms of future variations in academic achievement. With this challenge in our education systems, it is imperative for us to design transcending approaches, asking increasingly unpredictable questions, and no longer isolate ourselves in outdated and inefficient retrospective assessment cycles. In order to keep pace with the depth of expanding needs for high-quality medical education, previous studies reduced this gap through a unique combination of data mining and machine-learning methods, extending existing understanding of connections between past academic performance and educational outcomes [ 2 – 7 ]. Previous researches have also established a forecasting framework to circumvent some of the tensions of traditional regression-based methods [ 8 , 9 ]. While these statistical methods offered valuable foundational insights, they often failed to fully capture the complex nonlinear relationships inherent in educational datasets. Multiple algorithms have been seen the increasing application of machine learning techniques to address these limitations. Boruta algorithm can identify key predictors from multidimensional data, which define critical student’ profiles. Subsequently, machine learning algorithms can assess their predictive performance [ 10 – 13 ]. Integrating both Boruta and machine learning algorithms will address the fundamental limitations of previous methods by providing both predictive accuracy and practical interpretability [ 14 – 17 ]. Recently, SHAP (SHapley Additive exPlanations) not only outperform traditional statistical models but also resolve the “black-box” dilemma of complex algorithms [ 18 , 19 ]. Those advanced analytical techniques not only overcome the interference from multidimensional data but also provide novel methodologies for educational assessment. The developed tool enables medical institutions to shift from reactive to proactive student support, optimizing resource allocation and personalizing educational management. Our study will combine the Boruta algorithm and machine learning algorithms to define and evaluate notable predictors from the multidimensional data. By synthesizing conventional academic metrics with non-cognitive variables through advanced machine learning, our study established a new paradigm for medical education. The methodology and tools developed in this research will offer practical solutions for early identification of high-achieving students, targeted intervention strategies, and data-driven curriculum improvements, ultimately enhancing the quality and effectiveness of medical postgraduate training in China and beyond. Method Study objectives This study employed a semi-structured survey to collect data. Our study objectives included (1) comprehensive assessment of eight machine learning algorithms (2) interpreting predictive features by SHAP model and (3) development an applicable web calculator for predicting the students’ academic performance. Survey Instrument The “Your First Graduate Year” (YFGY) survey instrument was adapted from the widely used “Your First College Year” instrument [ 20 ], with careful cultural and pedagogical adaptations to ensure contextual relevance for China’s postgraduate education system. The final validated instrument was comprised of four areas after extensive pilot testing with 50 postgraduate students: (1) demographic data (e.g., age, gender, parental background data); (2) undergraduate academic profile; (3) postgraduate transition factors; and (4) multidimensional self-assessment factors. These objective academic factors were subjective cognitive factors that influence graduate students’ academic performance and could be systematically assessed through this soundly developed framework. Data Collection The questionnaire was collected during the orientation of postgraduate students in 2020 with the help of the university’s postgraduate administration. We utilized systematic data collection protocols to acquire complete and useful responses from participants at this important education transition point. The instrument was psychometrically sound, with Cronbach’s α = 0.884 indicating excellent internal reliability. The quality of the dataset was affirmed by the Kaiser-Meyer-Olkin test with a sampling adequacy measure of 0.915, and Bartlett’s test of sphericity (p < 0.001) confirmed a suitable framework for implementing further factor analysis. Collectively, these selected indicators established a strong movement toward the intended analysis. Also, all participants were offered a written informed consent form that outlined the voluntary nature of participation, specific research uses of the acquired data, and explicit permission for scientific publication. This careful ethical pathway not only ensured full and competent adherence to international research practice throughout the research, it also upheld the rights of the participants. Participants The study achieved remarkable engagement, collecting 1,133 complete questionnaires for a perfect initial response rate. After a thorough quality control screening, we noted that 42 questionnaires were invalidated due to invalid student identification numbers (n = 23), duplicate responses (n = 12), and inconsistent response patterns (n = 7). Subsequently, we built a final dataset of 1,091 strictly validated responses for a 96.3% validity rate. The detailed workflow was illustrated in Fig. 1. ------------------------------------- Insert Fig. 1 -------------------------------------- This carefully developed sample encompassed the entirety of the specialties covered in medical education comprising 43 academic specialties across seven colleges with disciplines under Clinical Medicine, Basic Medical Sciences, Traditional Chinese Medicine, Public Health, Nursing, Stomatology, and Biomedical Engineering. The colleges’ wide representation across a range of disciplinary areas helped ensure that our findings reflect the full range of medical postgraduate education in Chinese medical educational context, from clinical practice to biomedical research specialties. The broad representation of the sample and size (n = 1,091) allowed for credibility and real-world interpretations of the study’s results in regard to medical training pathways. Data Description The analytical dataset comprised 1,091 observations with 42 systematically categorized features (see Table 1). The variables collectively encompassed the complete educational pathway through four conceptually distinct and connected domains that represent a holistic, overlapping portrait of student potential. The first domain was Demographic Characteristics (7 variables) including gender, geographic origins, family structure, and select socioeconomic indicators, especially around parent education and parent occupation. Demographic Characteristics shaped the educational context students enter and influenced the degree of educational support surrounding them. The second domain was Undergraduate Academic Profile, which consisted of 17 variables, including undergraduate institution assignment (such as Institutional ranking), research output (thesis publications), language proficiency scores (CET-4/6, IELTS/TOEFL), any completed professional pre-requisite qualification (license to practice), behavioral information (study habits and hours and volunteerism/extracurriculars), and so on. Besides cumulative GPA, we paid more attention to academic preparation. The third domain was Postgraduate Transition Factors, which consisted of 9 variables. In order to fully capture candidate selection criteria of institution choice, we employed both practical (cost considerations, school reputation) and academic (major satisfaction, transfer (yes/no)) decisions. Finally, the Self-assessment Metrics, that were essentially nine variables, introduced not only a critical subjective dimension in terms of negotiating comparative competencies in problem-solving and learning capability, but also exploring plausible forecasts for future career trajectories. The combination of objective metrics and personal reflections presented an uncharacteristically holistic multi-dimensional perspective on the factors that enable success in medical postgraduate education. ------------------------------------- Insert Table 1 ------------------------------------- Dependent variable At the core of this study rested a product of careful deliberation, a dependent variable that reflected true academic achievement while overcoming the traditional limitations of GPA and grading. Acknowledging the variability within elective coursework over 43 medical specialties, we innovated by concentrating solely on mandatory core courses that addressed English, Medical Statistics, and Career Planning and computed a standardized “final GPA” as the arithmetic mean of these common requirements. This variable eliminated any potential biases from past elective selection while also providing a fair comparison metric across all disciplines. Performance classification followed the Southern Medical University “Regulations for Providing Early Graduation Recommendations of Outstanding Graduates” (Official Document No. 40 issued by Southern Medical University in 2016), with performance and Scholarship Recommendations. The 80-point GPA threshold created an important dichotomy of: high-achieving students ( \(\:\text{G}\text{P}\text{A}\ge\:80\) ) who were consistently excellent and average performing peers ( \(\:\text{G}\text{P}\text{A}<80\) ). This binary classification was much the same as the classification we would be reporting for analysis; it represented institutional standards to recognize identifiable exceptional talents. Based on our predictions should generate results that provide opportunities to examine organizations’ legitimate decision-making processes about academic standard performance. Machine Learning Models The Boruta algorithm, a complex wrapper method based on Random Forest (RF) classification logic, was used prior to building our model for more reliable feature selection. The Boruta algorithm uses an iterative comparison process by evaluating the relationships between variables systematically: each development of the model also generates faux, which are created by permuting the original variables at random. After it is calculated for each original and shadow feature, the original variables are retained if they demonstrate, on average, larger importance (Z-values) than the shadow features [ 21 ]. This method serves as an effective way to eliminate duplicates and/or uninformative feature variables to make certain that only features that offer statistically significant predictive value for GPA outcomes were being considered at the later stages of modeling. There were eight established ML algorithms that best reflect satisfying methodological paradigms as discussed in this paper: Flexible Discriminant Analysis (FDA), Linear Discriminant Analysis (LDA) Mixture Discriminant Analysis (MDA), Gradient Boosting Machine (GBM), Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost). The Boruta algorithm, a complex wrapper method based on RF classification logic, is used before building our model for more reliable feature selection. The Boruta algorithm uses an iterative comparison process by evaluating the relationships between variables systematically: each development of the model also generates shadow features, which are created by permuting the original variables at random. After it is calculated for each original and shadow feature, the original variables are retained if they demonstrate, on average, larger importance (Z-values) than the shadow features [ 22 , 23 ]. Model interpretability was enhanced through Shapley Additive Explanations (SHAP) analysis, where feature importance was quantified by absolute SHAP values-higher values indicate greater predictive influence [ 24 ]. The SHAP value for feature variable \(\:i\) was mathematically calculate by formula 1: In this case, M was the number of features, N was the set of all features, S was a subset of N that excluded features, if (S) was the model’s predicted value based only on the features in subset S, and the SHAP value of feature. A rise in GPA was indicated by a positive SHAP score, whilst a reduction in GPA was shown by a negative SHAP value. The contribution of each feature variable to GPA in ML models was examined using SHAP bar plots and summary plots; SHAP force graphs explained the outcomes of ML model predictions [ 25 , 26 ]. We randomly divided all qualified candidates into training and testing sets based on a 6:4 ratio. Using the testing set, all eight models were rigorously validated, and their performance was assessed using five important metrics: F1-score, accuracy, precision, recall, and the Area Under the ROC Curve (AUC). This multi-metric assessment framework was carefully chosen. The F1-score offered a reliable performance evaluation for situations involving imbalanced classification, the AUC measure assessed overall discriminative capacity, and the precision-recall balance informed intervention efficiency [ 27 , 28 ]. Thus, we made it easier clarify the model’s advantages and disadvantages in various prediction contexts; and enabled scalable implementation from individual student counseling to institutional policymaking. Statistical analysis strategy We analyzed the data to characterize the study population and assessed variable distributions. Categorical variables were summarized as frequencies and percentages, while continuous variables were compared between groups using independent samples t-tests. Statistical significance was set at p < 0.05 (two-tailed). For robust and reproducible results, we used SPSS 27.0 (IBM Corp.) for parametric testing and R (R Foundation) for advanced modeling. This dual-software approach ensured both precision and flexibility in the analysis. Results Comparison of baseline characteristics between the training and testing sets The study analyzed a cohort of 1,091 postgraduate’s students, randomly divided into training (n = 655, 60%) and testing (n = 436, 40%) sets using a carefully balanced 6:4 ratio. This partitioning strategy ensured sufficient sample size for model development while retaining adequate power for validation. Comparative analysis of baseline characteristics revealed no statistically significant differences between the datasets (all p-values > 0.05) These results, detailed in Table 2, demonstrated successful randomization and confirmed the absence of meaningful sampling bias. ------------------------------------- Insert Table 2 -------------------------------------- Feature variables screening based on the Boruta algorithm Feature selection utilized the Boruta algorithm which provided a novel tactic for determining which predictors are truly relevant. The basic flow of this algorithm was as follows: (1) creating shadow features by randomly permuting the original variable, (2) building an augmented dataset with shadow and actual features, and (3) utilizing Random Forest modeling and begin assessing variable importance. Each variable was classified into one of three groups based on the importance scores relative to the shadow features. First, if the important score for a feature was greater than the score of shadowMax, it could be classified as important; second, if it lied between the scores of shadowMean and shadowMax, it could undergo additional analysis; and third, if it lied below shadowMin, it could be passed off as noise. In our study, the Boruta package in R software was used to screen the relevant variables of GPA. The results were shown in Fig. 2. As the result showed, the important environmental variables with importance scores greater than shadowMax included 14 items: gender, undergraduate_GPA_rank, undergraduate_score_avreage, undergraduate_publication, CET-4, CET-6, fresh_graduate, highest_degree_expectation, highest_degree_expectation_SMU, contribute_to_country, pursue_higher_degree, make_money, confidence and career expectation. ------------------------------------- Insert Fig. 2 -------------------------------------- Construction and training machine learning models based on 14 feature variables Using the 14 predictive features identified as Fig. 2, we created and trained eight machine learning models to predict graduate academic performance. Overall comparisons highlighted that the XGBoost model achieved the best overall performance (Table 3), achieving an AUC of 0.744 (second best of all models), an accuracy of 0.828 (best), an F1-score of 902 (best), consistent precision (0.844) and recall (0.969) rates. Although models like the FDA performed competitively on certain metrics e.g., recall (0.966), but the FDA’s overall performance, as well as other models like SVM’s overall capability, appeared limited due to a lower AUC (0.666). XGBoost’s superior predictive performance could in part be attributed to its advanced abilities to model complex nonlinear interactions of features, as well as its ability to prevent overfitting via regularization, and its capability to accommodate high-dimensional educational data. These demonstrated advantages established XGBoost as the optimal algorithm for our predictive framework, warranting its selection for both practical implementation and in-depth result interpretation in educational contexts. The model’s strong performance across multiple evaluation dimensions suggested it was particularly well-suited for addressing the complexities inherent in predicting academic outcomes in medical graduate education. ------------------------------------- Insert Table 3 -------------------------------------- Verify the XGBoost model Figure 3 displayed the XGBoost model’s thorough performance evaluation, which used a variety of analytical viewpoints to gauge its predictive power. The XGBoost model achieved maximum net benefit within the clinically relevant threshold ranges of 0.3–0.7 for the test set (Fig. 3a) and 0.25–0.70 for the training set (Fig. 3b), which showed strong model performance across both training and testing datasets. This performance significantly outperformed the extreme strategies of classifying all cases as either “ordinary” or “excellent” demonstrating the model’s robust predictive value. With 326 (90.56%) correctly identified good performers, the test set’s confusion matrix analysis demonstrated exceptionally high classification accuracy (Fig. 3c). An Area Under Precision-Recall Curve of 0.906 (Fig. 3d), which indicated remarkable performance in differentiating academic attainment levels, was obtained through additional validation using precision-recall analysis. These consistent outcomes across all evaluation measures showed that the XGBoost model exhibited good generalizability from training to testing data and maintained outstanding predicted accuracy, indicating significant promise for real-world use in educational contexts. The model’s ability to maintain high performance across diverse evaluation frameworks underscored its reliability in medical graduate education. ------------------------------------- Insert Fig. 3 -------------------------------------- Model interpretation by SHAP analysis After a thorough analysis of several performance metrics, the XGBoost algorithm was determined to be the most effective method for building our student performance prediction model due to its high predictive accuracy. Although machine learning technique demonstrated good classification capabilities, its intrinsic complexity made it more difficult to understand compared to more straightforward models, such as logistic regression. Recognizing that educational stakeholders required actionable insights beyond simple predictions, we implemented SHAP model to elucidate the model’s decision-making processes and identify key determinants of academic performance. Three complimentary analytical techniques were used in our interpretability framework. In order to rank each predictive variable according to its relative contribution to GPA outcomes, we first created a global feature importance plot. The two most significant elements influencing students’ academic success were career expectations and undergraduate GPA rank. Inferior to above two variables, the remaining variables were ranked in descending order of importance as followed, contribute to country, undergraduate score average, fresh graduate, confidence, gender, CET-6, highest degree expectation, undergraduate publication, pursuit higher degree, make money, highest degree expectation SMU and CET-4, according to the critical feature analysis chart in Fig. 4a. In order to show the directional link between variable values and their influence on predictions, we secondly created feature summary graphs (SHAP beeswarm, Fig. 4b). The graph’s point represented an independent sample; the horizontal axis indicated the SHAP value. A positive SHAP value had a positive influence; one with a negative SHAP value had a negative impact. The purple to red color range showed that the SHAP value raised from low to high, indicating a progressive increase in the feature’s beneficial influence on the model output. Purple color which stood for the answer “no” denoted low eigenvalues, and red color which stood for the answer “yes” denoted high eigenvalues. The “two-end distribution” displayed by the SHAP values of each eigenvariable suggested that these 14 eigenvariables were capable of accurately predicting students’ GPA ratings. Higher job expectations and higher GPA ranks, for instance, typically caused the model to forecast positively. Overall, these characteristics had a systematic influence on the model’s prediction direction in addition to being numerically significant. Thirdly, we performed thorough comparisons of SHAP values between average-performing (true negative) and high-achieving (true positive) students based on selected five most important characteristics as Fig. 4a. The result (Fig. 5a) displayed the actual negative samples (the negative class that the model properly predicted) were shown in the figure as blue. True positivity, which the model accurately predicted to be positive, was indicated by orange. The findings demonstrated high interpretability and discrimination, as well as a substantial difference between the SHAP distributions of undergraduate_GPA_rank and career_expectation between positive and negative classes. To better elucidate variable contributions, we randomly selected a representative student case to demonstrate the interpretation results (Fig. 5b). This illustrative example revealed that positive prediction factors were undergraduate_ score_average (+ 0.406), CET-6 (+ 0.345), and make_money (+ 0.278). All three variables were positive contributions, representing that students’ average undergraduate grades and their English proficiency played a positive role in their final academic performance. The influence factors that had the greatest negative contribution to the prediction result was gender (-0.549), which represented that the student’s gender being male had a negative effect on the student’s final grade. These characteristic variables contributed to the output (f(x) = 0.821), which was less than the baseline value(E[f(X)] = 0), indicating an excellent level prediction. In addition to quantifying the predictive significance of each variable, this comprehensive analysis described the ways in which particular value ranged either favorably or unfavorably affected academic achievements. The resulting insights provided educators with a nuanced understanding of the complex interplay between student characteristics and academic outcomes, moving beyond binary classification to support data-informed intervention strategies. ------------------------------------- Insert Fig. 4 -------------------------------------- ------------------------------------- Insert Fig. 5 -------------------------------------- Development of a web-based application for predicting GPA The XGBoost binary classification model created in this study showed good predictive performance in the test set (n = 436), and we created a web-based application based on the anticipated risk of the final model (available at http://127.0.0.1:5963 ). There were 14 feature variables in the model, and there were 655 examples in the training set. Using a student A (real grade = 92.42) as an example (Fig. 5), the online-calculator predicted the probability of this student’s GPA ≥ 80 would reach 86.8%, which was significantly higher than the decision threshold of 50% (Fig. 3a), when the input parameters as Fig. 6 showed. The result of the student A showed that the online-calculator had a high level of confidence in the classification of the student A. Moreover, based on the explanatory graph of the contribution of SHAP features, it could be known confidence, contribute_to_country and female were the top three features contributing to her higher GPA, so as to provide ideas for subsequent cultivation for educators. ------------------------------------- Insert Fig. 6 -------------------------------------- Discussion Critical findings from our study The integration of big data analytics and machine learning had emerged as a transformative approach in educational research, particularly in predicting and analyzing student academic performance. This study utilized these advanced techniques to develop a robust GPA prediction framework for postgraduate’s students in a certain university. Through systematic comparison of eight machine learning algorithms using five evaluation metrics, the XGBoost model demonstrated superior predictive performance across all indicators, establishing its effectiveness for academic outcome forecasting. Through SHAP interpretability analysis, we found that the three most significant components were career_expectation, undergraduate_GPA_rank, and undergraduate_score_average. Therefore, the study offered important insights into the primary elements that influenced academic performance beyond prediction accuracy. The identified predictive patterns enabled a data-driven understanding of the multifaceted determinants of academic achievement, elucidated why certain students excel while others underperform. Meanwhile, the creation of an interactive online calculator made it easier to put these discoveries into practice by empowering educators to convert data-driven insights into workable solutions. These contributions provided a reproducible framework for organizations looking to improve student outcomes through evidence-based practices, advancing both theoretical knowledge and real-world applications in educational analytics. Critical influences in predicting GPA Our established model could provide reliable the data to similar student populations in real-world academic settings. Our comprehensive investigation examined four critical dimensions of influence: (1) individual demographic characteristics, (2) undergraduate academic profile, (3) postgraduate transition factors, and (4) self-assessment metrics. Demographic Characteristics Academic performance was found to be significantly predicted by gender, with female students showing better GPAs than male students (Table 2). This finding was consistent with previous research that links structural and sociocultural factors to gender differences in academic performance. For instance, Lievens et al. [ 29 ] identified curriculum design and gender imbalances in student cohorts as potential contributors. In the context of Chinese higher education, societal expectations and traditional gender roles might further encourage female students to success academically, thereby securing better employment opportunities in a competitive job market [ 30 ]. These findings were corroborated by Cuddy et al. [ 31 ], who stated that female individuals demonstrated superior performance in clinical knowledge compared to males [ 32 ]. Females generally exhibited stronger self-discipline and better learning habits (e.g., timely completion of assignments and higher classroom participation), which might serve as primary contributing factors [ 33 ]. Undergraduate Academic Profile Undergraduate GPA ranking and average scores were strongly correlated with postgraduate academic performance as our findings, reinforced by Yusuf et al. [ 34 – 36 ]. Pervious studied showed that those from higher-status institutions were more likely to progress to postgraduate study and were also more likely to be from the more socioeconomically advantaged groups [ 37 , 38 ]. Interestingly, the reputation of undergraduate institutions exhibited minimal influence on graduate GPAs, challenging the conventional assumption that graduates from elite institutions performing better in postgraduate programs. Also, we found individual academic merit such as GPA and research experience provide a more reliable indicator of future success. These insights suggested that admissions policies should prioritize the academic ability of applicants over institutional prestige to foster a fairer and more meritocratic selection process. Postgraduate Transition Factors Academic performance (GPA) was significantly influenced by graduation status (current-year vs. non-current-year graduates), career expectation, and enthusiasm for further study (Fig. 4). Our results were supported by Nabizadeh et.al [ 39 ], which indicted career expectations in particular were the most significant factors for GPA. Students who had clear professional aspirations showed higher levels of self-efficacy and academic perseverance [ 40 ]. Moreover, the mediating function of career aspirations in promoting university satisfaction in turn strengthened academic commitment, emphasized by Kikuchi et al. [ 41 ]. Graduation status also played a role in influencing the GPAs in postgraduate students, and our results showed that current-year graduates demonstrated significantly higher GPAs than non-current-year graduates. This discrepancy may stem from the latter group’s potential distractions (e.g., work or family obligations) and the challenges of readapting to academic life after an academic break [ 42 ]. Self-assessment Metrics High self-assessment might contribute to improved academic performance. Students’ positive attitudes toward learning and high level of self-confidence had a big impact on their good academic performance (Fig. 4). Higher GPAs were more likely to be attained by those who had higher levels of self-efficacy that they could overcome academic obstacles [ 43 ]. Furthermore, students who were motivated primarily by financial incentives showed lower levels of academic engagement than those who were motivated by altruistic goals, such as making a positive contribution to society [ 44 ]. These results highlighted how crucial it was to promote intrinsic motivation and a good view of oneself in learning environments. Our research findings were supported by Liu et al. [ 45 ], which stated that students who exhibited high levels of intrinsic motivation demonstrated superior performance compared to those with lower levels of intrinsic motivation. Methodological Contributions and Practical Applications Our model demonstrated superior performance in predicting the academic performance. Different from the traditional models, which mostly relied on linear predictors (such as undergraduate GPA and GRE scores [ 46 ]), our study used machine learning approaches to capture nonlinear interactions among factors. The XGBoost algorithm could efficiently manage high-dimensional data, intricate feature interactions, and adjust hyperparameter in order to reduce the danger of overfitting. Our online GPA calculator has increased the usefulness of research in the real world, enabling educators to identify students’ academic performance early and allocate educational resources rationally. Through the integration of interpretability frameworks such as SHAP with MLs, our study offered practical insights for enhancing educational results and guiding policy decisions. These analytical findings offered substantial practical value for educational practitioners and policymakers. Specifically, the results provided actionable insights for: (1) developing targeted programs for high-achieving students, (2) optimizing admission criteria to identify high-potential candidates, and (3) designing curriculum enhancements that addressed the most influential success factors. The successful application of SHAP analysis in this context also established an important methodological precedent, demonstrating how advanced interpretability techniques could bridge the gap between complex machine learning outputs and practical educational decision-making. This approach not only validated the model’s predictive mechanisms, but also created a transparent framework for translating algorithmic findings into evidence-based pedagogical strategies. Consequently, our study enhanced the practical utility of predictive analytics in higher education management. Limitations and future work Our study yielded several educationally significant findings, yet several limitations should be noted. The restricted sample size and lack of diverse demographic variables such as marital status might affect the model’s generalizability and accuracy. Additionally, the current online calculator required ongoing updates to incorporate new predictive features and ensured compatibility with evolving educational data standards. Future work should focus on expanding the dataset with more representative student populations, integrating additional influential variables into the model, and continuously maintaining the web platform to enhance functionality and user accessibility. Conclusion This research made a significant contribution by combining interpretable SHAP analysis with machine learning model (XGBoost), which improved predicted accuracy and offered useful insights into the factors that influence student achievement. Based on our model, online calculator was developed to assist educators in early identification of high-achieving students. Our study offered both data support and case-based assurance for implementing personalized training approaches in graduate education, thereby optimizing the allocation of educational resources and improving institutional outcomes. Abbreviations GPA Grade point average ML Machine learning SMU Southern Medical University EDM Educational Data Mining XGBoost Extreme gradient boosting FDA Fisher discriminant analysis LDA Linear discriminant analysis MDA Multiple discriminant analysis GBM Gradient boosting machine LR Logistic Regression RF Regression Forest SVM Support vector machine AUC Area under the receiver SHAP Shapley additive explanation Declarations Funding None. Author information Authors and Affiliations School of Public Health, Southern Medical University, Guangzhou, Guangdong, China Wenyi Lai & Guan Ying & Kailiang Shen Office of Academic Affairs, Southern Medical University, Guangzhou, Guangdong, China Jinna Lin Graduate School, Southern Medical University, Guangzhou, Guangdong, China Zhihai Ling Contributions LWY and GY conceived and designed the study. LZH, SKL, and LWY contributed to the acquisition, analysis, and interpretation of data. LZH and LJN were involved in investigation and data collection. LWY drafted the manuscript. GY revised subsequent drafts. All authors reviewed and approved the final manuscript. Corresponding author Correspondence to Ying Guan. Competing interests The authors declare no competing interests. Ethics declarations Ethics approval and consent to participate Ethics approval was obtained from the Biomedical Ethics Committee of Southern Medical University and was approved by each participating institution. Written informed consent for publication was obtained from all participants. All research was conducted in adherence with the Declaration of Helsinki. Clinical trial number: not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author [GY] on reasonable request. References Education statistics: Ministry of Education of the People’s Republic of China Web site. https://kaoyan.eol.cn/e_ky/zt/report/2025/content03.html Picton, A.; Greenfield, S.; Parry, J. Why Do Students Struggle in Their First Year of Medical School? A Qualitative Study of Student Voices. BMC Medical Education 2022, 22 (1), 100. https://doi.org/10.1186/s12909-022-03158-4. Puddey, I. B.; Mercer, A.; Carr, S. E. Relative Progress and Academic Performance of Graduate vs Undergraduate Entrants to an Australian Medical School. BMC Medical Education 2019, 19 (1), 159. https://doi.org/10.1186/s12909-019-1584-0. Li, J.; Thompson, R.; Shulruf, B. Struggling with Strugglers: Using Data from Selection Tools for Early Identification of Medical Students at Risk of Failure. BMC Medical Education 2019, 19 (1), 415. https://doi.org/10.1186/s12909-019-1860-z. AHMADY, S.; KHAJEALI, N.; SHARIFI, F.; MIRMOGHTADAEI, Z. S. Factors Related to Academic Failure in Preclinical Medical Education: A Systematic Review. J Adv Med Educ Prof 2019, 7 (2), 74–85. https://doi.org/10.30476/JAMP.2019.44711. Shukri, A. K.; Mubarak, A. S. Factors of Academic Success Among Undergraduate Medical Students in Taif University, Saudi Arabia: A Cross-Sectional Study. 2019. Monteverde-Suárez, D.; González-Flores, P.; Santos-Solórzano, R.; García-Minjares, M.; Zavala-Sierra, I.; de la Luz, V. L.; Sánchez-Mendiola, M. Predicting Students’ Academic Progress and Related Attributes in First-Year Medical Students: An Analysis with Artificial Neural Networks and Naïve Bayes. BMC Medical Education 2024, 24 (1), 74. https://doi.org/10.1186/s12909-023-04918-6. Esmat, T. A.; Pitts, J. D. Predicting Success in an Undergraduate Exercise Science Program Using Science-Based Admission Courses. Advances in Physiology Education 2020, 44 (2), 138–144. https://doi.org/10.1152/advan.00130.2019. Huberts, L. C. E.; Schoonhoven, Marit; and Does, R. J. M. M. Multilevel Process Monitoring: A Case Study to Predict Student Success or Failure. Journal of Quality Technology 2022, 54 (2), 127–143. https://doi.org/10.1080/00224065.2020.1828008. Bir, D.; Ahn, B. Factors Predicting Students’ Persistence and Academic Success in an Aerospace Engineering Program*. O. Embarak. Apply Machine Learning Algorithms to Predict At-Risk Students to Admission Period. In 2020 Seventh International Conference on Information Technology Trends (ITT); 2020; pp 190–195. https://doi.org/10.1109/ITT51279.2020.9320878. Zhang Feng; Chen Jingjing. A Review of Learning Data Characteristics Suitable for Student Performance Prediction. Software engineering, 2023, 26 (10), 1-4. https://doi.org/10.19644/j.cnki.issn2096-1472.2023.010.001. Merchant, A. A. H.; Afzal, N.; Rahim, K. A.; Shah, S. A.; Jamal, W. Z.; Rahim, A.; Bakhshi, S. K.; Shaikh, N. Q.; Noorali, A. A.; Mahmood, S. B. Z.; Akbarali, L.; Tariq, M.; Khan, S.; Haider, A. H. Application to Achievement: Association between Pre-Admission Factors, Admission Scores, and Medical Students’ Performance. BMC Medical Education 2025, 25 (1), 223. https://doi.org/10.1186/s12909-025-06800-z. V. L. Uskov; J. P. Bakken; A. Byerly; A. Shah. Machine Learning-Based Predictive Analytics of Student Academic Performance in STEM Education. In 2019 IEEE Global Engineering Education Conference (EDUCON); 2019; pp 1370–1376. https://doi.org/10.1109/EDUCON.2019.8725237. Feng, J.; Shen, C.; Zhang, D.; Yang, W.; Xu, G. Development and Validation of a Nomogram to Predict Hemiplegic Shoulder Pain in Patients With Stroke: A Retrospective Cohort Study. Archives of Rehabilitation Research and Clinical Translation 2022, 4 (3), 100213. https://doi.org/10.1016/j.arrct.2022.100213. Kim, Y. H.; Jung, S. J.; Yang, E. J.; Paik, N. J. Clinical and Sonographic Risk Factors for Hemiplegic Shoulder Pain: A Longitudinal Observational Study. J Rehabil Med 2014, 46 (1), 81–87. https://doi.org/10.2340/16501977-1238. Yue, S.; Li, S.; Huang, X.; Liu, J.; Hou, X.; Zhao, Y.; Niu, D.; Wang, Y.; Tan, W.; Wu, J. Machine Learning for the Prediction of Acute Kidney Injury in Patients with Sepsis. J Transl Med 2022, 20 (1), 215. https://doi.org/10.1186/s12967-022-03364-0. Wang, K.; Tian, J.; Zheng, C.; Yang, H.; Ren, J.; Liu, Y.; Han, Q.; Zhang, Y. Interpretable Prediction of 3-Year All-Cause Mortality in Patients with Heart Failure Caused by Coronary Heart Disease Based on Machine Learning and SHAP. Comput Biol Med 2021, 137, 104813. https://doi.org/10.1016/j.compbiomed.2021.104813. Shu, L.; Yan, H.; Wu, Y.; Yan, T.; Yang, L.; Zhang, S.; Chen, Z.; Liao, Q.; Yang, L.; Xiao, B.; Ye, M.; Lv, S.; Wu, M.; Zhu, X.; Hu, P. Explainable Machine Learning in Outcome Prediction of High-Grade Aneurysmal Subarachnoid Hemorrhage. Aging (Albany NY) 2024, 16 (5), 4654–4669. https://doi.org/10.18632/aging.205621. Shen Jiahao, Guan Ying, Ou Chunquan, Tan Xiaoxue, Yu Shuang, Ling Zhihai, Chen Zheng. Analysis of Influencing Factors of Medical Master’s Degree Course Examination Scores Based on Structural Equation Model [J] Chinese Health Statistics,2022,39(05):695-698. Ma Chao, Huang Huaguo, Tian Xin, Liu Bingjie, Wen Kunjian, Wang Pengjie. Research on the Estimation Method of Single Wood Volume Based on Backpack LiDAR [J]. Remote Sensing Technology and Application, 2022, 37 (05): 1071-1083. Cui Tingting, An Xuelian, Sun Deliang, Chen Dongsheng, Zhu Youchen. Evaluation Model of landslide Susceptibility Based on SHAP interpretable Machine Learning [J]. Journal of Chengdu University of Technology (Natural Science Edition), 2025, 52 (01): 153-172. DeBerard, M. S.; Spielmans, G. I.; Julka, D. L. Predictors of Academic Achievement and Retention among College Freshmen: A Longitudinal Study. College Student Journal 2004, 38 (1), 66–80. Ye, Z.; An, S.; Gao, Y.; Xie, E.; Zhao, X.; Guo, Z.; Li, Y.; Shen, N.; Zheng, J. Association between the Triglyceride Glucose Index and In-Hospital and 1-Year Mortality in Patients with Chronic Kidney Disease and Coronary Artery Disease in the Intensive Care Unit. Cardiovasc Diabetol 2023, 22 (1), 110. https://doi.org/10.1186/s12933-023-01843-2. Farzipour, A.; Elmi, R.; Nasiri, H. Detection of Monkeypox Cases Based on Symptoms Using XGBoost and Shapley Additive Explanations Methods. Diagnostics (Basel) 2023, 13 (14), 2391. https://doi.org/10.3390/diagnostics13142391. Shu, L.; Yan, H.; Wu, Y.; Yan, T.; Yang, L.; Zhang, S.; Chen, Z.; Liao, Q.; Yang, L.; Xiao, B.; Ye, M.; Lv, S.; Wu, M.; Zhu, X.; Hu, P. Explainable Machine Learning in Outcome Prediction of High-Grade Aneurysmal Subarachnoid Hemorrhage. Aging (Albany NY) 2024, 16 (5), 4654–4669. https://doi.org/10.18632/aging.205621. Zhang Ze, Chu Zhe. Research on Student Performance Prediction and SHAP Feature Analysis Based on XGBoost Informationization Research. 2024; 50 (03): 34-40. Zan Ziyi, Yue Weifeng, Zhao Hangzheng, Cao Changming, Hu Jingdan, Hu Litang, Xu Yang, Chen Aiping. Construction of Agricultural Water Consumption Prediction Model Based on Machine Learning and SHAP Algorithm [J]. Water Saving Irrigation, 2024, (12): 102-110. Lievens, F.; Buyse, T.; Sackett, P. R. Retest Effects in Operational Selection Settings: Development and Test of a Framework. Personnel Psychology 2005, 58 (4), 981–1007. https://doi.org/10.1111/j.1744-6570.2005.00713.x. Cuddy, M. M.; Swanson, D. B.; Clauser, B. E. A Multilevel Analysis of Examinee Gender and USMLE Step 1 Performance. Acad Med 2008, 83 (10 Suppl), S58-62. https://doi.org/10.1097/ACM.0b013e318183cd65. Cuddy, M. M.; Swanson, D. B.; Clauser, B. E. A Multilevel Analysis of the Relationships between Examinee Gender and United States Medical Licensing Exam (USMLE) Step 2 CK Content Area Performance. Acad Med 2007, 82 (10 Suppl), S89-93. https://doi.org/10.1097/ACM.0b013e3181400379. Qiao Lianquan, Li Lingling. The Influence of Parents’ Educational Attainment on the Self-Positioning of Graduate Students: A Case Study of Doctoral Students at Xiamen University [J]. Hubei College Entrance Examination, 2011, (30): 7-11. Obregon-Cuesta, A. I.; Mínguez-Mínguez, L. A.; León-Del-Barco, B.; Mendo-Lázaro, S.; Fernández-Solana, J.; González-Bernal, J. J.; González-Santos, J. Bullying in Adolescents: Differences between Gender and School Year and Relationship with Academic Performance. Int J Environ Res Public Health 2022, 19 (15), 9301. https://doi.org/10.3390/ijerph19159301. Yusuf, M.; Adigun, J. The Influence of School Sex, Location and Type on Students’ Academic Performance. International Journal of Educational Sciences 2010, 2, 81–85. https://doi.org/10.1080/09751122.2010.11889992. Titus, A.; Adu, E. School Location and Gender as Correlates of Students' Academic Achievement in Economics. Goel, A. D.; Akarte, S. V.; Agrawal, S. P.; Yadav, V. Longitudinal Assessment of Depression, Stress, and Burnout in Medical Students. J Neurosci Rural Pract 2016, 7 (4), 493–498. https://doi.org/10.4103/0976-3147.188625. Scott, P. (2020). Access to Postgraduate Study: Representation and Destinations. Scottish Government - Commissioner for Fair Access. Wakeling, P., & Laurison, D. (2017). Are Postgraduate Qualifications the ‘New Frontier of Social Mobility’? British Journal of Sociology,68(3), 533–555. Nabizadeh, S.; Hajian, S.; Sheikhan, Z.; Rafiei, F. Prediction of Academic Achievement Based on Learning Strategies and Outcome Expectations among Medical Students. BMC Med Educ 2019, 19 (1), 99. https://doi.org/10.1186/s12909-019-1527-9. Perrone, K.; Tschopp, M.; Snyder, E.; Boo, J.; Hyatt, C. A Longitudinal Examination of Career Expectations and Outcomes of Academically Talented Students 10- and 20-Years Post—High School Graduation. Journal of Career Development - J CAREER DEVELOPMENT 2010, 36, 291–309. https://doi.org/10.1177/0894845309359347. Kikuchi, A.; Kawamoto, R.; Abe, M.; Ninomiya, D.; Tokumoto, Y.; Kumagi, T. Ambiguous Motivations in Medical School Applicants: A Retrospective Study from Japan. Med Educ Online 2025, 30 (1), 2467487. https://doi.org/10.1080/10872981.2025.2467487. Wang Hui. Investigation and Research on the Admission Motivations of Master’s Degree Students with Interdisciplinary Background in Education [Master’s Thesis] Shanghai Normal University 2016. Zhu Xiaowen, Ding Wenjun, Zhang Mingliang. Research on the Current Situation and Enhancement Strategies of Positive Psychological Capital of Students in Police Colleges: Taking the Students of the 2017 Grade of Shandong Police College as Samples [J] Journal of Public Security (Journal of Zhejiang Police College), 2018, (06): 90-94. Mateos-González, J. L.; Wakeling, P. Exploring Socioeconomic Inequalities and Access to Elite Postgraduate Education among English Graduates. Higher Education 2022, 83 (3), 673–694. https://doi.org/10.1007/s10734-021-00693-9. Liu, Y.; Hau, K.-T.; Zheng, X. Does Instrumental Motivation Help Students with Low Intrinsic Motivation? Comparison between Western andConfucian Students. International journal of psychology: Journal international de psychologie 2020, 55 (2), 182–191. https://doi.org/10.1002/ijop.12563. Hyun, M. S. Examining the Validity of the GRE General Test Scores and Undergraduate GPA for Predicting Success in Graduate School at a Large Racially and Ethnically Diverse Public University in Southeast Florida. FIU Electronic Theses and Dissertations 2012. https://doi.org/10.25148/etd.FI12121002. Tables Table 1. Description of postgraduate feature variables Variable Coding scheme Demographic Characteristics gender 1=female;2=male province 1=Guangdong;2=non-Guangdong only_child a 1=yes;2=no household_registration 1=urban;2=rural highest_degree_parents 1=illiterate;2=primary school;3=middle school;4=high school;5=bachelor degree;6=master/Ph.D father_occupation 1=government office, national enterprise, and institution;2=worker;3=farmer;4=businessman;5=service practitioner;6=soldier;7=production and transportation personnel;8=other mother_occupation 1=government office, national enterprise, and institution;2=worker;3=farmer;4=businessman;5=service practitioner;6=soldier;7=production and transportation personnel;8=other Undergraduate Academic Profile undergraduate_GPA_rank 1=top 5%;2=top 10%;3=top 20-50%;4=below 50% undergraduate_colleges 1=985 university;2=211 university;3=SMU and others undergraduate_score_average b 1=60-69;2=70-79;3=80-89;4=≥90 undergraduate_publication 1=0;2=≥1 first_author_publication 1=yes;2=no SCI_publication 1=yes;2=no CET-4 1=pass;2=fail to pass CET-6 1=pass;2=fail to pass IELTS_or_TOEFL 1=pass;2=fail to pass physician’s_medical_license c 1=yes;2=no need punishment 1=yes;2=no frequency_activities 1=often;2=occasionally;3=never negative_behaviors 1=often;2=occasionally;3=never study_behaviors 1=often;2=occasionally;3=never self-improvement 1=often;2=occasionally;3=never recreation_duration 1=≤7 hours;2=>7 hours study_duration 1=≤7 hours;2=>7 hours Postgraduate Transition Factors reputation_SMU 1=very important;2=somewhat important;3=insignificance reasonable_tuition 1=very important;2=somewhat important;3=insignificance satisfied_major 1=very satisfied;2=somewhat satisfied;3=a little dissatisfied;4=very dissatisfied transfer_school d 1=yes;2=no fresh_graduate 1=yes;2=no degree_category 1=academic master;2=professional master highest_degree_expectation 1=none;2=master’s degree;3=doctor’s degree highest_degree_expectation_SMU 1=none;2=master’s degree;3=doctor’s degree career_expectation 1=doctor;2=college teacher;3=scientific researchers;4=employee of the enterprise;5=administrator;6=other Self-assessment metrics contribute_to_country 1=absolutely necessary;2=very important;3=somewhat important;4=insignificance artistic_achievement 1=absolutely necessary;2=very important;3=somewhat important;4=insignificance recognized_in_work 1=absolutely necessary;2=very important;3=somewhat important;4=insignificance pursue_higher_degree 1=very important;2=somewhat important;3=insignificance make_money 1=very important;2=somewhat important;3=insignificance confidence e 1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower problem-solving_ability f 1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower learning_technical_ability g 1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower continue_study 1=stand a good chance;2=have a chance;3=a very small chance;4=have no chance a : It refers to the only child born to a couple. b : The calculation method of this variable is “weighted average score = (∑ [(a certain subject score * the same subject credit)])/total credits”. c : It refers to the qualification certificate uniformly issued by the National Health Commission after passing the national unified practicing physician qualification examination and practicing assistant physician qualification examination. It is a necessary certificate for practicing physicians in China. d : It means that during the postgraduate admission process, if a candidate meets the Re-examination conditions but cannot participate in the Re-examination at the first-choice institution or fails the Re-examination，the candidate can choose to transfer to other institutions. e : It refers to how an individual perceives their own level of self-confidence compared with the average level of their peers. f : It refers to how an individual perceives their own problem-solving ability compared with the average level of their peers. g : It refers to how an individual perceives their own learning technical ability compared with the average level of their peers. Table 2. Comparison of master’s student data between training set and test set Variable training set （ n=655 ） testing set （ n=436 ） P-value Demographic characteristics gender 1.168 0.280 male 242(36.3) 168(39.5) female 425(63.7) 257(60.5) province 0.217 0.642 Guangdong 273(40.9) 180(42.4) non-Guangdong 394(59.1) 245(57.6) only child 225(33.7) 141(33.2) 0.036 0.849 household_registration 0.506 0.477 urban 349(52.3) 213(50.1) rural 318(47.7) 212(49.9) highest_degree_parents 10.993 0.052 illiterate 2(0.3) 0(0.0) primary 52(7.8) 51(12.0) Middle school 250(37.5) 150(35.3) high school 196(29.4) 133(31.3) bachelor degree 142(21.3) 84(19.8) master/Ph.D 25(3.7) 7(1.6) father_occupation 10.652 0.155 government/ national enterprise and institution 105(15.7) 60(14.1) worker 99(14.8) 85(20.0) farmer 155(23.2) 97(22.8) businessman 79(11.8) 53(12.5) service practitioner 20(3.0) 19(4.5) soldier 3(0.4) 0(0.0) production and transportation personnel 24(3.6) 16(3.8) other 182(27.3) 95(22.4) mother_occupation 6.755 0.344 government/national enterprise and institution 102(15.3) 50(11.8) worker 77(11.5) 64(15.1) farmer 155(23.2) 102(24.0) businessman 46(6.9) 33(7.8) service practitioner 30(4.5) 25(5.9) soldier 0(0.0) 0(0.0) production and transportation personnel 30(4.5) 25(5.9) other 250(37.5) 147(34.6) Characteristics of undergraduate period undergraduate_GPA_rank 2.348 0.503 top 5% 107(16.0) 68(16) top 10% 173(25.9) 94(22.1) top 20-50% 318(47.7) 213(50.1) below 50% 69(10.3) 50(11.8) undergraduate_colleges 1.763 0.414 985 university 11(1.6) 12(2.8) 211 university 35(5.2) 23(5.4) SMU and others 621(93.1) 390(91.8) undergraduate_score_average 6.833 0.077 60-69 18(2.7) 16(3.8) 70-79 239(35.8) 165(38.8) 80-89 401(60.1) 231(54.4) ≥90 9(1.3) 13(3.1) undergraduate_publication ≥1 107(16.0) 66(15.5) 0.051 0.821 first author publication paper 53(7.9) 23(5.4) 3.598 1.165 published SCI paper during undergraduate 29(4.3) 18(4.2) 0.052 0.974 passed CET-4 659(98.8) 415(97.6) 2.131 0.144 passed CET-6 530(79.5) 328(77.2) 0.804 0.370 passed IELTS_or_TOEFL 35(5.2) 8(1.9) 7.771 0.005 acquired physician’s_medical_license 32(4.8) 17(4.0) 2.433 0.295 no punishment 665(99.7) 425(100) 1.277 0.259 often frequency_activities 89(13.3) 55(12.9) 0.327 0.849 often negative_behaviors 2(0.3) 0(0.0) 1.464 0.481 often study_behaviors 286(42.9) 190(44.7) 1.448 0.485 often self-improvement 142(21.3) 92(21.6) 0.918 0.632 recreation duration>7 hours 9(1.3) 8(1.9) 2.497 0.869 study duration>7 hours 87(13.0) 47(11.1) 4.159 0.655 Characteristics as a Postgraduate candidate SMU’s reputation is very important 326(48.9) 221(52.0) 1.437 0.487 reasonable SMU’s tuition is very important 187(28.0) 146(34.4) 4.888 0.087 satisfied_major 644(96.5) 409(96.2) 0.897 0.826 first-choice candidate 451(67.6) 286(67.3) 0.012 0.912 fresh graduate 492(73.8) 312(73.4) 0.017 0.898 degree_category 0.152 0.696 academic master 492(73.8) 318(74.8) professional master 175(26.2) 107(25.2) highest_degree_expectation 2.241 0.326 none 36(5.4) 17(4.0) master’s degree 138(20.7) 78(18.4) doctor’s degree 493(73.9) 330(77.6) highest_degree_expectation_SMU 0.690 0.708 none 22(3.3) 18(4.2) master’s degree 253(37.9) 157(36.9) doctor’s degree 392(58.8) 250(58.8) career_expectation 5.484 0.360 doctor 263(39.4) 161(37.9) college teacher 108(16.2) 70(16.5) scientific researchers 137(20.5) 99(23.3) employee of the enterprise 35(5.2) 30(7.1) administrator 47(7.0) 19(4.5) other 77(11.5) 46(10.8) Self-cognitive assessments. more confident than others 206(30.8) 144(33.9) 1.446 0.836 a better problem solver than others 184(27.6) 148(34.8) 8.342 0.039 a better learner of technology than others 113(16.9) 80(34.8) 1.184 0.881 a good chance to further study 286(42.8) 203(47.8) 5.569 0.135 absolutely necessary to make a contribution to the country 71(10.6) 61(14.4) 6.640 0.084 absolutely necessary to make achievements in art 22(3.3) 23(5.4) 7.940 0.047 absolutely necessary to be recognized in your field 172(25.8) 116(27.3) 3.877 0.275 get a higher degree is very important 463(69.4) 299(70.4) 0.824 0.662 make money is very important 508(76.2) 329(77.4) 2.612 0.271 Table 3. Predictive performance of the eight machine learning techniques Model AUC Accuracy Precision Recall F1-score FDA 0.705 0.825 0.844 0.966 0.901 LDA 0.724 0.816 0.851 0.941 0.894 MDA 0.710 0.791 0.845 0.913 0.878 GBM 0.735 0.823 0.838 0.972 0.900 LR 0.737 0.832 0.848 0.969 0.905 RF 0.699 0.818 0.823 0.992 0.900 SVM 0.666 0.823 0.832 0.983 0.901 XGBoost 0.744 0.828 0.844 0.969 0.902 Note: AUC, the value of the area under the three classification curves; FDA, Flexible Discriminant Analysis; LDA, Linear Discriminant Analysis; MDA, Mixture Discriminant Analysis; GBM, Gradient Boosting Machine; LR, Logistic Regression; RF, Random Forest; SVM, Support Vector Machine; XGBoost, Extreme Gradient Boosting. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 10 Feb, 2026 Read the published version in BMC Medical Education → Version 1 posted Editorial decision: Revision requested 03 Oct, 2025 Reviews received at journal 24 Sep, 2025 Reviewers agreed at journal 24 Sep, 2025 Reviewers agreed at journal 22 Sep, 2025 Reviews received at journal 19 Sep, 2025 Reviewers agreed at journal 09 Sep, 2025 Reviewers agreed at journal 06 Sep, 2025 Reviewers invited by journal 04 Sep, 2025 Editor assigned by journal 02 Sep, 2025 Editor invited by journal 18 Aug, 2025 Submission checks completed at journal 16 Aug, 2025 First submitted to journal 16 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7186245","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":512707881,"identity":"11fd37ba-6c07-4801-bfaf-2421383a7544","order_by":0,"name":"Wenyi Lai","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wenyi","middleName":"","lastName":"Lai","suffix":""},{"id":512707882,"identity":"415f0ad2-f420-4a3c-b04e-42d93a4c4542","order_by":1,"name":"Jinna Lin","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jinna","middleName":"","lastName":"Lin","suffix":""},{"id":512707883,"identity":"ece32748-70f9-4f4e-a29c-54bdb59c58e0","order_by":2,"name":"Kailiang Shen","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kailiang","middleName":"","lastName":"Shen","suffix":""},{"id":512707884,"identity":"f7fed7a8-8a16-49bc-9bb8-c24002115492","order_by":3,"name":"Zhihai Ling","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhihai","middleName":"","lastName":"Ling","suffix":""},{"id":512707885,"identity":"342da7f8-d98d-4222-b454-36d0d001d45e","order_by":4,"name":"Ying Guan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYNACAwYGfgiLmQQtkg2kaQHpOkCsFvmI5GePeQru2G0+fzpNgqHCOrGB/ewBvFoMb6SZG/MYPEvediN3mwTDmfTEBp68BPxaZiSYSfMYHE42u8G7TYKx7XBigwSPAQEt6d/AWoz7zwK1/CNCi7xEDtgWOwMGoMMYG4jQYsDzpkxyjsHhBIkbuZstEo6lG7fx5BCwpT19m8SbP4ft+fvPbrzxocZatp/9DAFbDjAwMPEwMCQ2gHgJQMyGVz3IFqBSxh8MDPaEFI6CUTAKRsEIBgBxkEOaX0m34AAAAABJRU5ErkJggg==","orcid":"","institution":"Southern Medical University","correspondingAuthor":true,"prefix":"","firstName":"Ying","middleName":"","lastName":"Guan","suffix":""}],"badges":[],"createdAt":"2025-07-22 11:08:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7186245/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7186245/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12909-026-08741-7","type":"published","date":"2026-02-10T15:58:12+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":91070772,"identity":"73062b06-4d67-4c5f-b834-a54282947166","added_by":"auto","created_at":"2025-09-11 10:46:15","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":54443,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the research process\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/c5bb785187a566cf4a061329.png"},{"id":91069804,"identity":"75836672-c928-4c4a-81fb-aa44af69c8ba","added_by":"auto","created_at":"2025-09-11 10:38:16","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":30988,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance ranking of the XGBoost model using the Boruta algorithm\u003c/p\u003e\n\u003cp\u003eThis bar chart shows the relative importance of all input features ranked by their contribution to the model based on the Boruta feature selection algorithm. Higher values indicate stronger influence on GPA prediction.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/905f3464651ced91373e6700.png"},{"id":91069803,"identity":"8bae6d0f-8850-4a46-8d8f-dcd1ca1329e0","added_by":"auto","created_at":"2025-09-11 10:38:16","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":356686,"visible":true,"origin":"","legend":"\u003cp\u003eEvaluation of the XGBoost model’s performance in predicting postgraduate GPA\u003c/p\u003e\n\u003cp\u003e(a) Decision curve for the training set, indicating the net benefit across different threshold probabilities compared to “treat all” and “treat none” strategies. (b) Decision curve for the testing set, demonstrating the model’s clinical utility and stability on unseen data. (c) Confusion matrix showing the classification results between predicted and actual GPA categories (“0” = Ordinary, “1” = Outstanding). (d) Precision-Recall (PR) curve with an Area Under the PR Curve (AUPRC) of 0.906, indicating excellent classification performance.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/88c908c4a681f63410cda870.jpeg"},{"id":91069785,"identity":"f0fc3404-2044-4285-b48a-6743da5ba0d9","added_by":"auto","created_at":"2025-09-11 10:38:15","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":135799,"visible":true,"origin":"","legend":"\u003cp\u003eGlobal SHAP explanation of GPA prediction based on XGBoost\u003c/p\u003e\n\u003cp\u003e(a) Bar chart showing the mean absolute SHAP value of each feature, indicating its overall importance in the XGBoost model. (b) SHAP beeswarm plot showing the impact of individual feature values on the model output. Each dot represents a student; red indicates a high feature value, purple indicates a low feature value. Positive SHAP values push the prediction toward a higher GPA, while negative values push it lower.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/bff09eba2311acf4c1eabd18.png"},{"id":91069788,"identity":"cae9c161-b812-4f52-be99-33f599f62162","added_by":"auto","created_at":"2025-09-11 10:38:15","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":200819,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP values in different prediction classes and for an individual high-GPA student\u003c/p\u003e\n\u003cp\u003e(a) Distribution of SHAP values for five key features among true positive and true negative samples. Boxplots and density plots show how feature impacts differ by prediction class, reflecting discriminative power. Interpretation of the Prediction Results of the XGBoost Model Based on the SHAP Waterfall Plot. (b)SHAP value bar chart for a representative top-performing student, illustrating how features positively contributed to the GPA prediction.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/5880b78b26f5c98dcaf6435a.png"},{"id":91069796,"identity":"2986f710-fc32-4a78-af33-9561f2d4ce9f","added_by":"auto","created_at":"2025-09-11 10:38:16","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":350007,"visible":true,"origin":"","legend":"\u003cp\u003eWeb-based GPA prediction tool developed using the XGBoost model\u003c/p\u003e\n\u003cp\u003eThis interface demonstrates the GPA prediction system for postgraduate students. The user inputs values for key predictors (e.g., undergraduate GPA rank, CET-4/6, career expectation, etc.), and the system outputs a predicted GPA classification (≥80 or \u0026lt;80), along with the corresponding probabilities. SHAP-based explanations are provided to interpret the individual prediction.\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/78eeffe527bc92a0312452a7.jpeg"},{"id":102785585,"identity":"228f3c48-e10c-4303-8054-05eeccdb0fbd","added_by":"auto","created_at":"2026-02-16 16:08:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2581460,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7186245/v1/077dc2bb-3f54-4975-87f0-da88e916a797.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Enhancing Graduate Education Assessment: A Machine Learning Approach to GPA Prediction for Medical Students","fulltext":[{"header":"Background","content":"\u003cp\u003eRecently Chinese medical postgraduate education has changed extraordinarily in recent years with enrolments reaching 156,000 students in 2023, representing 12% of postgraduate admissions nationally [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. The current two-step evaluation system (standardized assessment combined with comprehensive re-assessment) is effective for assessing theoretical knowledge and clinical competencies, but lacks predictive capabilities in terms of future variations in academic achievement. With this challenge in our education systems, it is imperative for us to design transcending approaches, asking increasingly unpredictable questions, and no longer isolate ourselves in outdated and inefficient retrospective assessment cycles.\u003c/p\u003e\u003cp\u003eIn order to keep pace with the depth of expanding needs for high-quality medical education, previous studies reduced this gap through a unique combination of data mining and machine-learning methods, extending existing understanding of connections between past academic performance and educational outcomes [\u003cspan additionalcitationids=\"CR3 CR4 CR5 CR6\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e–\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Previous researches have also established a forecasting framework to circumvent some of the tensions of traditional regression-based methods [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. While these statistical methods offered valuable foundational insights, they often failed to fully capture the complex nonlinear relationships inherent in educational datasets. Multiple algorithms have been seen the increasing application of machine learning techniques to address these limitations. Boruta algorithm can identify key predictors from multidimensional data, which define critical student’ profiles. Subsequently, machine learning algorithms can assess their predictive performance [\u003cspan additionalcitationids=\"CR11 CR12\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e–\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Integrating both Boruta and machine learning algorithms will address the fundamental limitations of previous methods by providing both predictive accuracy and practical interpretability [\u003cspan additionalcitationids=\"CR15 CR16\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e–\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Recently, SHAP (SHapley Additive exPlanations) not only outperform traditional statistical models but also resolve the “black-box” dilemma of complex algorithms [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Those advanced analytical techniques not only overcome the interference from multidimensional data but also provide novel methodologies for educational assessment. The developed tool enables medical institutions to shift from reactive to proactive student support, optimizing resource allocation and personalizing educational management.\u003c/p\u003e\u003cp\u003eOur study will combine the Boruta algorithm and machine learning algorithms to define and evaluate notable predictors from the multidimensional data. By synthesizing conventional academic metrics with non-cognitive variables through advanced machine learning, our study established a new paradigm for medical education. The methodology and tools developed in this research will offer practical solutions for early identification of high-achieving students, targeted intervention strategies, and data-driven curriculum improvements, ultimately enhancing the quality and effectiveness of medical postgraduate training in China and beyond.\u003c/p\u003e"},{"header":"Method","content":"\u003cp\u003eStudy objectives\u003c/p\u003e\u003cp\u003eThis study employed a semi-structured survey to collect data. Our study objectives included (1) comprehensive assessment of eight machine learning algorithms (2) interpreting predictive features by SHAP model and (3) development an applicable web calculator for predicting the students’ academic performance.\u003c/p\u003e\u003cp\u003eSurvey Instrument\u003c/p\u003e\u003cp\u003eThe “Your First Graduate Year” (YFGY) survey instrument was adapted from the widely used “Your First College Year” instrument [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], with careful cultural and pedagogical adaptations to ensure contextual relevance for China’s postgraduate education system. The final validated instrument was comprised of four areas after extensive pilot testing with 50 postgraduate students: (1) demographic data (e.g., age, gender, parental background data); (2) undergraduate academic profile; (3) postgraduate transition factors; and (4) multidimensional self-assessment factors. These objective academic factors were subjective cognitive factors that influence graduate students’ academic performance and could be systematically assessed through this soundly developed framework.\u003c/p\u003e\u003cp\u003eData Collection\u003c/p\u003e\u003cp\u003eThe questionnaire was collected during the orientation of postgraduate students in 2020 with the help of the university’s postgraduate administration. We utilized systematic data collection protocols to acquire complete and useful responses from participants at this important education transition point. The instrument was psychometrically sound, with Cronbach’s α = 0.884 indicating excellent internal reliability. The quality of the dataset was affirmed by the Kaiser-Meyer-Olkin test with a sampling adequacy measure of 0.915, and Bartlett’s test of sphericity (p \u0026lt; 0.001) confirmed a suitable framework for implementing further factor analysis. Collectively, these selected indicators established a strong movement toward the intended analysis. Also, all participants were offered a written informed consent form that outlined the voluntary nature of participation, specific research uses of the acquired data, and explicit permission for scientific publication. This careful ethical pathway not only ensured full and competent adherence to international research practice throughout the research, it also upheld the rights of the participants.\u003c/p\u003e\u003cp\u003eParticipants\u003c/p\u003e\u003cp\u003eThe study achieved remarkable engagement, collecting 1,133 complete questionnaires for a perfect initial response rate. After a thorough quality control screening, we noted that 42 questionnaires were invalidated due to invalid student identification numbers (n = 23), duplicate responses (n = 12), and inconsistent response patterns (n = 7). Subsequently, we built a final dataset of 1,091 strictly validated responses for a 96.3% validity rate. The detailed workflow was illustrated in Fig.\u0026nbsp;1.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;1\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eThis carefully developed sample encompassed the entirety of the specialties covered in medical education comprising 43 academic specialties across seven colleges with disciplines under Clinical Medicine, Basic Medical Sciences, Traditional Chinese Medicine, Public Health, Nursing, Stomatology, and Biomedical Engineering. The colleges’ wide representation across a range of disciplinary areas helped ensure that our findings reflect the full range of medical postgraduate education in Chinese medical educational context, from clinical practice to biomedical research specialties. The broad representation of the sample and size (n = 1,091) allowed for credibility and real-world interpretations of the study’s results in regard to medical training pathways.\u003c/p\u003e\u003cp\u003eData Description\u003c/p\u003e\u003cp\u003eThe analytical dataset comprised 1,091 observations with 42 systematically categorized features (see Table\u0026nbsp;1). The variables collectively encompassed the complete educational pathway through four conceptually distinct and connected domains that represent a holistic, overlapping portrait of student potential.\u003c/p\u003e\u003cp\u003eThe first domain was Demographic Characteristics (7 variables) including gender, geographic origins, family structure, and select socioeconomic indicators, especially around parent education and parent occupation. Demographic Characteristics shaped the educational context students enter and influenced the degree of educational support surrounding them.\u003c/p\u003e\u003cp\u003eThe second domain was Undergraduate Academic Profile, which consisted of 17 variables, including undergraduate institution assignment (such as Institutional ranking), research output (thesis publications), language proficiency scores (CET-4/6, IELTS/TOEFL), any completed professional pre-requisite qualification (license to practice), behavioral information (study habits and hours and volunteerism/extracurriculars), and so on. Besides cumulative GPA, we paid more attention to academic preparation.\u003c/p\u003e\u003cp\u003eThe third domain was Postgraduate Transition Factors, which consisted of 9 variables. In order to fully capture candidate selection criteria of institution choice, we employed both practical (cost considerations, school reputation) and academic (major satisfaction, transfer (yes/no)) decisions.\u003c/p\u003e\u003cp\u003eFinally, the Self-assessment Metrics, that were essentially nine variables, introduced not only a critical subjective dimension in terms of negotiating comparative competencies in problem-solving and learning capability, but also exploring plausible forecasts for future career trajectories. The combination of objective metrics and personal reflections presented an uncharacteristically holistic multi-dimensional perspective on the factors that enable success in medical postgraduate education.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Table\u0026nbsp;1\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eDependent variable\u003c/p\u003e\u003cp\u003eAt the core of this study rested a product of careful deliberation, a dependent variable that reflected true academic achievement while overcoming the traditional limitations of GPA and grading. Acknowledging the variability within elective coursework over 43 medical specialties, we innovated by concentrating solely on mandatory core courses that addressed English, Medical Statistics, and Career Planning and computed a standardized “final GPA” as the arithmetic mean of these common requirements. This variable eliminated any potential biases from past elective selection while also providing a fair comparison metric across all disciplines.\u003c/p\u003e\u003cp\u003ePerformance classification followed the Southern Medical University “Regulations for Providing Early Graduation Recommendations of Outstanding Graduates” (Official Document No. 40 issued by Southern Medical University in 2016), with performance and Scholarship Recommendations. The 80-point GPA threshold created an important dichotomy of: high-achieving students (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{G}\\text{P}\\text{A}\\ge\\:80\\)\u003c/span\u003e\u003c/span\u003e) who were consistently excellent and average performing peers (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{G}\\text{P}\\text{A}\u0026lt;80\\)\u003c/span\u003e\u003c/span\u003e). This binary classification was much the same as the classification we would be reporting for analysis; it represented institutional standards to recognize identifiable exceptional talents. Based on our predictions should generate results that provide opportunities to examine organizations’ legitimate decision-making processes about academic standard performance.\u003c/p\u003e\u003ch3\u003eMachine Learning Models\u003c/h3\u003e\u003cp\u003eThe Boruta algorithm, a complex wrapper method based on Random Forest (RF) classification logic, was used prior to building our model for more reliable feature selection. The Boruta algorithm uses an iterative comparison process by evaluating the relationships between variables systematically: each development of the model also generates faux, which are created by permuting the original variables at random. After it is calculated for each original and shadow feature, the original variables are retained if they demonstrate, on average, larger importance (Z-values) than the shadow features [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. This method serves as an effective way to eliminate duplicates and/or uninformative feature variables to make certain that only features that offer statistically significant predictive value for GPA outcomes were being considered at the later stages of modeling.\u003c/p\u003e\u003cp\u003eThere were eight established ML algorithms that best reflect satisfying methodological paradigms as discussed in this paper: Flexible Discriminant Analysis (FDA), Linear Discriminant Analysis (LDA) Mixture Discriminant Analysis (MDA), Gradient Boosting Machine (GBM), Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost). The Boruta algorithm, a complex wrapper method based on RF classification logic, is used before building our model for more reliable feature selection. The Boruta algorithm uses an iterative comparison process by evaluating the relationships between variables systematically: each development of the model also generates shadow features, which are created by permuting the original variables at random. After it is calculated for each original and shadow feature, the original variables are retained if they demonstrate, on average, larger importance (Z-values) than the shadow features [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eModel interpretability was enhanced through Shapley Additive Explanations (SHAP) analysis, where feature importance was quantified by absolute SHAP values-higher values indicate greater predictive influence [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The SHAP value for feature variable \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e was mathematically calculate by formula 1:\u003c/p\u003e\u003cp\u003e\u003cimg src=\"https://myfiles.space/user_files/83062_751fab6dfaef2446/83062_custom_files/img1757586691.png\" width=\"535\" height=\"60.7387\" style=\"width: 535px; height: 60.7387px;\"\u003e\u003c/p\u003e\u003cp\u003eIn this case, M was the number of features, N was the set of all features, S was a subset of N that excluded features, if (S) was the model’s predicted value based only on the features in subset S, and the SHAP value of feature.\u003c/p\u003e\u003cp\u003eA rise in GPA was indicated by a positive SHAP score, whilst a reduction in GPA was shown by a negative SHAP value. The contribution of each feature variable to GPA in ML models was examined using SHAP bar plots and summary plots; SHAP force graphs explained the outcomes of ML model predictions [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eWe randomly divided all qualified candidates into training and testing sets based on a 6:4 ratio. Using the testing set, all eight models were rigorously validated, and their performance was assessed using five important metrics: F1-score, accuracy, precision, recall, and the Area Under the ROC Curve (AUC). This multi-metric assessment framework was carefully chosen. The F1-score offered a reliable performance evaluation for situations involving imbalanced classification, the AUC measure assessed overall discriminative capacity, and the precision-recall balance informed intervention efficiency [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Thus, we made it easier clarify the model’s advantages and disadvantages in various prediction contexts; and enabled scalable implementation from individual student counseling to institutional policymaking.\u003c/p\u003e\u003cp\u003eStatistical analysis strategy\u003c/p\u003e\u003cp\u003eWe analyzed the data to characterize the study population and assessed variable distributions. Categorical variables were summarized as frequencies and percentages, while continuous variables were compared between groups using independent samples t-tests. Statistical significance was set at p \u0026lt; 0.05 (two-tailed). For robust and reproducible results, we used SPSS 27.0 (IBM Corp.) for parametric testing and R (R Foundation) for advanced modeling. This dual-software approach ensured both precision and flexibility in the analysis.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eComparison of baseline characteristics between the training and testing sets\u003c/p\u003e\u003cp\u003eThe study analyzed a cohort of 1,091 postgraduate\u0026rsquo;s students, randomly divided into training (n\u0026thinsp;=\u0026thinsp;655, 60%) and testing (n\u0026thinsp;=\u0026thinsp;436, 40%) sets using a carefully balanced 6:4 ratio. This partitioning strategy ensured sufficient sample size for model development while retaining adequate power for validation. Comparative analysis of baseline characteristics revealed no statistically significant differences between the datasets (all p-values\u0026thinsp;\u0026gt;\u0026thinsp;0.05) These results, detailed in Table\u0026nbsp;2, demonstrated successful randomization and confirmed the absence of meaningful sampling bias.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Table\u0026nbsp;2\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eFeature variables screening based on the Boruta algorithm\u003c/p\u003e\u003cp\u003eFeature selection utilized the Boruta algorithm which provided a novel tactic for determining which predictors are truly relevant. The basic flow of this algorithm was as follows: (1) creating shadow features by randomly permuting the original variable, (2) building an augmented dataset with shadow and actual features, and (3) utilizing Random Forest modeling and begin assessing variable importance. Each variable was classified into one of three groups based on the importance scores relative to the shadow features. First, if the important score for a feature was greater than the score of shadowMax, it could be classified as important; second, if it lied between the scores of shadowMean and shadowMax, it could undergo additional analysis; and third, if it lied below shadowMin, it could be passed off as noise.\u003c/p\u003e\u003cp\u003eIn our study, the Boruta package in R software was used to screen the relevant variables of GPA. The results were shown in Fig.\u0026nbsp;2. As the result showed, the important environmental variables with importance scores greater than shadowMax included 14 items: gender, undergraduate_GPA_rank, undergraduate_score_avreage, undergraduate_publication, CET-4, CET-6, fresh_graduate, highest_degree_expectation, highest_degree_expectation_SMU, contribute_to_country, pursue_higher_degree, make_money, confidence and career expectation.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;2\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eConstruction and training machine learning models based on 14 feature variables\u003c/p\u003e\u003cp\u003eUsing the 14 predictive features identified as Fig.\u0026nbsp;2, we created and trained eight machine learning models to predict graduate academic performance. Overall comparisons highlighted that the XGBoost model achieved the best overall performance (Table\u0026nbsp;3), achieving an AUC of 0.744 (second best of all models), an accuracy of 0.828 (best), an F1-score of 902 (best), consistent precision (0.844) and recall (0.969) rates. Although models like the FDA performed competitively on certain metrics e.g., recall (0.966), but the FDA\u0026rsquo;s overall performance, as well as other models like SVM\u0026rsquo;s overall capability, appeared limited due to a lower AUC (0.666). XGBoost\u0026rsquo;s superior predictive performance could in part be attributed to its advanced abilities to model complex nonlinear interactions of features, as well as its ability to prevent overfitting via regularization, and its capability to accommodate high-dimensional educational data. These demonstrated advantages established XGBoost as the optimal algorithm for our predictive framework, warranting its selection for both practical implementation and in-depth result interpretation in educational contexts. The model\u0026rsquo;s strong performance across multiple evaluation dimensions suggested it was particularly well-suited for addressing the complexities inherent in predicting academic outcomes in medical graduate education.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Table\u0026nbsp;3\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eVerify the XGBoost model\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;3 displayed the XGBoost model\u0026rsquo;s thorough performance evaluation, which used a variety of analytical viewpoints to gauge its predictive power. The XGBoost model achieved maximum net benefit within the clinically relevant threshold ranges of 0.3\u0026ndash;0.7 for the test set (Fig.\u0026nbsp;3a) and 0.25\u0026ndash;0.70 for the training set (Fig.\u0026nbsp;3b), which showed strong model performance across both training and testing datasets. This performance significantly outperformed the extreme strategies of classifying all cases as either \u0026ldquo;ordinary\u0026rdquo; or \u0026ldquo;excellent\u0026rdquo; demonstrating the model\u0026rsquo;s robust predictive value. With 326 (90.56%) correctly identified good performers, the test set\u0026rsquo;s confusion matrix analysis demonstrated exceptionally high classification accuracy (Fig.\u0026nbsp;3c). An Area Under Precision-Recall Curve of 0.906 (Fig.\u0026nbsp;3d), which indicated remarkable performance in differentiating academic attainment levels, was obtained through additional validation using precision-recall analysis. These consistent outcomes across all evaluation measures showed that the XGBoost model exhibited good generalizability from training to testing data and maintained outstanding predicted accuracy, indicating significant promise for real-world use in educational contexts. The model\u0026rsquo;s ability to maintain high performance across diverse evaluation frameworks underscored its reliability in medical graduate education.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;3\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eModel interpretation by SHAP analysis\u003c/p\u003e\u003cp\u003eAfter a thorough analysis of several performance metrics, the XGBoost algorithm was determined to be the most effective method for building our student performance prediction model due to its high predictive accuracy. Although machine learning technique demonstrated good classification capabilities, its intrinsic complexity made it more difficult to understand compared to more straightforward models, such as logistic regression. Recognizing that educational stakeholders required actionable insights beyond simple predictions, we implemented SHAP model to elucidate the model\u0026rsquo;s decision-making processes and identify key determinants of academic performance.\u003c/p\u003e\u003cp\u003eThree complimentary analytical techniques were used in our interpretability framework. In order to rank each predictive variable according to its relative contribution to GPA outcomes, we first created a global feature importance plot. The two most significant elements influencing students\u0026rsquo; academic success were career expectations and undergraduate GPA rank. Inferior to above two variables, the remaining variables were ranked in descending order of importance as followed, contribute to country, undergraduate score average, fresh graduate, confidence, gender, CET-6, highest degree expectation, undergraduate publication, pursuit higher degree, make money, highest degree expectation SMU and CET-4, according to the critical feature analysis chart in Fig.\u0026nbsp;4a.\u003c/p\u003e\u003cp\u003eIn order to show the directional link between variable values and their influence on predictions, we secondly created feature summary graphs (SHAP beeswarm, Fig.\u0026nbsp;4b). The graph\u0026rsquo;s point represented an independent sample; the horizontal axis indicated the SHAP value. A positive SHAP value had a positive influence; one with a negative SHAP value had a negative impact. The purple to red color range showed that the SHAP value raised from low to high, indicating a progressive increase in the feature\u0026rsquo;s beneficial influence on the model output. Purple color which stood for the answer \u0026ldquo;no\u0026rdquo; denoted low eigenvalues, and red color which stood for the answer \u0026ldquo;yes\u0026rdquo; denoted high eigenvalues. The \u0026ldquo;two-end distribution\u0026rdquo; displayed by the SHAP values of each eigenvariable suggested that these 14 eigenvariables were capable of accurately predicting students\u0026rsquo; GPA ratings. Higher job expectations and higher GPA ranks, for instance, typically caused the model to forecast positively. Overall, these characteristics had a systematic influence on the model\u0026rsquo;s prediction direction in addition to being numerically significant.\u003c/p\u003e\u003cp\u003eThirdly, we performed thorough comparisons of SHAP values between average-performing (true negative) and high-achieving (true positive) students based on selected five most important characteristics as Fig.\u0026nbsp;4a. The result (Fig.\u0026nbsp;5a) displayed the actual negative samples (the negative class that the model properly predicted) were shown in the figure as blue. True positivity, which the model accurately predicted to be positive, was indicated by orange. The findings demonstrated high interpretability and discrimination, as well as a substantial difference between the SHAP distributions of undergraduate_GPA_rank and career_expectation between positive and negative classes.\u003c/p\u003e\u003cp\u003eTo better elucidate variable contributions, we randomly selected a representative student case to demonstrate the interpretation results (Fig.\u0026nbsp;5b). This illustrative example revealed that positive prediction factors were undergraduate_ score_average (+\u0026thinsp;0.406), CET-6 (+\u0026thinsp;0.345), and make_money (+\u0026thinsp;0.278). All three variables were positive contributions, representing that students\u0026rsquo; average undergraduate grades and their English proficiency played a positive role in their final academic performance. The influence factors that had the greatest negative contribution to the prediction result was gender (-0.549), which represented that the student\u0026rsquo;s gender being male had a negative effect on the student\u0026rsquo;s final grade. These characteristic variables contributed to the output (f(x)\u0026thinsp;\u003cb\u003e=\u003c/b\u003e\u0026thinsp;0.821), which was less than the baseline value(E[f(X)]\u0026thinsp;=\u0026thinsp;0), indicating an excellent level prediction.\u003c/p\u003e\u003cp\u003eIn addition to quantifying the predictive significance of each variable, this comprehensive analysis described the ways in which particular value ranged either favorably or unfavorably affected academic achievements. The resulting insights provided educators with a nuanced understanding of the complex interplay between student characteristics and academic outcomes, moving beyond binary classification to support data-informed intervention strategies.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;4\u003c/p\u003e\u003cp\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;5\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e\u003cp\u003eDevelopment of a web-based application for predicting GPA\u003c/p\u003e\u003cp\u003eThe XGBoost binary classification model created in this study showed good predictive performance in the test set (n\u0026thinsp;=\u0026thinsp;436), and we created a web-based application based on the anticipated risk of the final model (available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://127.0.0.1:5963\u003c/span\u003e\u003cspan address=\"http://127.0.0.1:5963\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). There were 14 feature variables in the model, and there were 655 examples in the training set. Using a student A (real grade\u0026thinsp;=\u0026thinsp;92.42) as an example (Fig.\u0026nbsp;5), the online-calculator predicted the probability of this student\u0026rsquo;s GPA\u0026thinsp;\u0026ge;\u0026thinsp;80 would reach 86.8%, which was significantly higher than the decision threshold of 50% (Fig.\u0026nbsp;3a), when the input parameters as Fig.\u0026nbsp;6 showed. The result of the student A showed that the online-calculator had a high level of confidence in the classification of the student A. Moreover, based on the explanatory graph of the contribution of SHAP features, it could be known confidence, contribute_to_country and female were the top three features contributing to her higher GPA, so as to provide ideas for subsequent cultivation for educators.\u003c/p\u003e\u003cp\u003e-------------------------------------\u003c/p\u003e\u003cp\u003eInsert Fig.\u0026nbsp;6\u003c/p\u003e\u003cp\u003e--------------------------------------\u003c/p\u003e"},{"header":"Discussion","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003eCritical findings from our study\u003c/h2\u003e\u003cp\u003eThe integration of big data analytics and machine learning had emerged as a transformative approach in educational research, particularly in predicting and analyzing student academic performance. This study utilized these advanced techniques to develop a robust GPA prediction framework for postgraduate\u0026rsquo;s students in a certain university. Through systematic comparison of eight machine learning algorithms using five evaluation metrics, the XGBoost model demonstrated superior predictive performance across all indicators, establishing its effectiveness for academic outcome forecasting. Through SHAP interpretability analysis, we found that the three most significant components were career_expectation, undergraduate_GPA_rank, and undergraduate_score_average. Therefore, the study offered important insights into the primary elements that influenced academic performance beyond prediction accuracy. The identified predictive patterns enabled a data-driven understanding of the multifaceted determinants of academic achievement, elucidated why certain students excel while others underperform. Meanwhile, the creation of an interactive online calculator made it easier to put these discoveries into practice by empowering educators to convert data-driven insights into workable solutions. These contributions provided a reproducible framework for organizations looking to improve student outcomes through evidence-based practices, advancing both theoretical knowledge and real-world applications in educational analytics.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eCritical influences in predicting GPA\u003c/h3\u003e\n\u003cp\u003eOur established model could provide reliable the data to similar student populations in real-world academic settings. Our comprehensive investigation examined four critical dimensions of influence: (1) individual demographic characteristics, (2) undergraduate academic profile, (3) postgraduate transition factors, and (4) self-assessment metrics.\u003c/p\u003e\n\u003ch3\u003eDemographic Characteristics\u003c/h3\u003e\n\u003cp\u003eAcademic performance was found to be significantly predicted by gender, with female students showing better GPAs than male students (Table\u0026nbsp;2). This finding was consistent with previous research that links structural and sociocultural factors to gender differences in academic performance. For instance, Lievens et al. [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] identified curriculum design and gender imbalances in student cohorts as potential contributors. In the context of Chinese higher education, societal expectations and traditional gender roles might further encourage female students to success academically, thereby securing better employment opportunities in a competitive job market [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. These findings were corroborated by Cuddy et al. [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], who stated that female individuals demonstrated superior performance in clinical knowledge compared to males [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Females generally exhibited stronger self-discipline and better learning habits (e.g., timely completion of assignments and higher classroom participation), which might serve as primary contributing factors [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e].\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eUndergraduate Academic Profile\u003c/h2\u003e\u003cp\u003eUndergraduate GPA ranking and average scores were strongly correlated with postgraduate academic performance as our findings, reinforced by Yusuf et al. [\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Pervious studied showed that those from higher-status institutions were more likely to progress to postgraduate study and were also more likely to be from the more socioeconomically advantaged groups [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Interestingly, the reputation of undergraduate institutions exhibited minimal influence on graduate GPAs, challenging the conventional assumption that graduates from elite institutions performing better in postgraduate programs. Also, we found individual academic merit such as GPA and research experience provide a more reliable indicator of future success. These insights suggested that admissions policies should prioritize the academic ability of applicants over institutional prestige to foster a fairer and more meritocratic selection process.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003ePostgraduate Transition Factors\u003c/h3\u003e\n\u003cp\u003eAcademic performance (GPA) was significantly influenced by graduation status (current-year vs. non-current-year graduates), career expectation, and enthusiasm for further study (Fig.\u0026nbsp;4). Our results were supported by Nabizadeh et.al [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e], which indicted career expectations in particular were the most significant factors for GPA. Students who had clear professional aspirations showed higher levels of self-efficacy and academic perseverance [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Moreover, the mediating function of career aspirations in promoting university satisfaction in turn strengthened academic commitment, emphasized by Kikuchi et al. [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Graduation status also played a role in influencing the GPAs in postgraduate students, and our results showed that current-year graduates demonstrated significantly higher GPAs than non-current-year graduates. This discrepancy may stem from the latter group\u0026rsquo;s potential distractions (e.g., work or family obligations) and the challenges of readapting to academic life after an academic break [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eSelf-assessment Metrics\u003c/h3\u003e\n\u003cp\u003eHigh self-assessment might contribute to improved academic performance. Students\u0026rsquo; positive attitudes toward learning and high level of self-confidence had a big impact on their good academic performance (Fig.\u0026nbsp;4). Higher GPAs were more likely to be attained by those who had higher levels of self-efficacy that they could overcome academic obstacles [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Furthermore, students who were motivated primarily by financial incentives showed lower levels of academic engagement than those who were motivated by altruistic goals, such as making a positive contribution to society [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. These results highlighted how crucial it was to promote intrinsic motivation and a good view of oneself in learning environments. Our research findings were supported by Liu et al. [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e], which stated that students who exhibited high levels of intrinsic motivation demonstrated superior performance compared to those with lower levels of intrinsic motivation.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eMethodological Contributions and Practical Applications\u003c/h2\u003e\u003cp\u003eOur model demonstrated superior performance in predicting the academic performance. Different from the traditional models, which mostly relied on linear predictors (such as undergraduate GPA and GRE scores [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]), our study used machine learning approaches to capture nonlinear interactions among factors. The XGBoost algorithm could efficiently manage high-dimensional data, intricate feature interactions, and adjust hyperparameter in order to reduce the danger of overfitting. Our online GPA calculator has increased the usefulness of research in the real world, enabling educators to identify students\u0026rsquo; academic performance early and allocate educational resources rationally. Through the integration of interpretability frameworks such as SHAP with MLs, our study offered practical insights for enhancing educational results and guiding policy decisions.\u003c/p\u003e\u003cp\u003eThese analytical findings offered substantial practical value for educational practitioners and policymakers. Specifically, the results provided actionable insights for: (1) developing targeted programs for high-achieving students, (2) optimizing admission criteria to identify high-potential candidates, and (3) designing curriculum enhancements that addressed the most influential success factors. The successful application of SHAP analysis in this context also established an important methodological precedent, demonstrating how advanced interpretability techniques could bridge the gap between complex machine learning outputs and practical educational decision-making. This approach not only validated the model\u0026rsquo;s predictive mechanisms, but also created a transparent framework for translating algorithmic findings into evidence-based pedagogical strategies. Consequently, our study enhanced the practical utility of predictive analytics in higher education management.\u003c/p\u003e\u003cp\u003eLimitations and future work\u003c/p\u003e\u003cp\u003eOur study yielded several educationally significant findings, yet several limitations should be noted. The restricted sample size and lack of diverse demographic variables such as marital status might affect the model\u0026rsquo;s generalizability and accuracy. Additionally, the current online calculator required ongoing updates to incorporate new predictive features and ensured compatibility with evolving educational data standards. Future work should focus on expanding the dataset with more representative student populations, integrating additional influential variables into the model, and continuously maintaining the web platform to enhance functionality and user accessibility.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis research made a significant contribution by combining interpretable SHAP analysis with machine learning model (XGBoost), which improved predicted accuracy and offered useful insights into the factors that influence student achievement. Based on our model, online calculator was developed to assist educators in early identification of high-achieving students. Our study offered both data support and case-based assurance for implementing personalized training approaches in graduate education, thereby optimizing the allocation of educational resources and improving institutional outcomes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGPA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGrade point average\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eML\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMachine learning\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSMU\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSouthern Medical University\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eEDM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eEducational Data Mining\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eXGBoost\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eExtreme gradient boosting\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eFDA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eFisher discriminant analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eLDA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLinear discriminant analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eMDA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMultiple discriminant analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGBM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGradient boosting machine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eLR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLogistic Regression\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eRF\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRegression Forest\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSVM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSupport vector machine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eArea under the receiver\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSHAP\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eShapley additive explanation\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003cp\u003eAuthor information\u003c/p\u003e\n\u003cp\u003eAuthors and Affiliations\u003c/p\u003e\n\u003cp\u003eSchool of Public Health, Southern Medical University, Guangzhou, Guangdong, China\u003c/p\u003e\n\u003cp\u003eWenyi Lai \u0026amp; Guan Ying \u0026amp; Kailiang Shen\u003c/p\u003e\n\u003cp\u003eOffice of Academic Affairs, Southern Medical University, Guangzhou, Guangdong, China\u003c/p\u003e\n\u003cp\u003eJinna Lin\u003c/p\u003e\n\u003cp\u003eGraduate School, Southern Medical University, Guangzhou, Guangdong, China\u003c/p\u003e\n\u003cp\u003eZhihai Ling\u003c/p\u003e\n\u003cp\u003eContributions\u003c/p\u003e\n\u003cp\u003eLWY and GY conceived and designed the study. LZH, SKL, and LWY contributed to the acquisition, analysis, and interpretation of data. LZH and LJN were involved in investigation and data collection. LWY drafted the manuscript. GY revised subsequent drafts. All authors reviewed and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eCorresponding author\u003c/p\u003e\n\u003cp\u003eCorrespondence to Ying Guan.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eEthics declarations\u003c/p\u003e\n\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eEthics approval was obtained from the Biomedical Ethics Committee of Southern Medical University and was approved by each participating institution. Written informed consent for publication was obtained from all participants.\u0026nbsp;All research was conducted in adherence with the Declaration of Helsinki. Clinical trial number: not applicable.\u003c/p\u003e\n\u003cp\u003eConsent for publication\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author [GY] on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eEducation statistics: Ministry of Education of the People\u0026rsquo;s Republic of China Web site. https://kaoyan.eol.cn/e_ky/zt/report/2025/content03.html\u003c/li\u003e\n\u003cli\u003ePicton, A.; Greenfield, S.; Parry, J. Why Do Students Struggle in Their First Year of Medical School? A Qualitative Study of Student Voices. BMC Medical Education 2022, 22 (1), 100. https://doi.org/10.1186/s12909-022-03158-4.\u003c/li\u003e\n\u003cli\u003ePuddey, I. B.; Mercer, A.; Carr, S. E. Relative Progress and Academic Performance of Graduate vs Undergraduate Entrants to an Australian Medical School. BMC Medical Education 2019, 19 (1), 159. https://doi.org/10.1186/s12909-019-1584-0.\u003c/li\u003e\n\u003cli\u003eLi, J.; Thompson, R.; Shulruf, B. Struggling with Strugglers: Using Data from Selection Tools for Early Identification of Medical Students at Risk of Failure. BMC Medical Education 2019, 19 (1), 415. https://doi.org/10.1186/s12909-019-1860-z.\u003c/li\u003e\n\u003cli\u003eAHMADY, S.; KHAJEALI, N.; SHARIFI, F.; MIRMOGHTADAEI, Z. S. Factors Related to Academic Failure in Preclinical Medical Education: A Systematic Review. J Adv Med Educ Prof 2019, 7 (2), 74\u0026ndash;85. https://doi.org/10.30476/JAMP.2019.44711.\u003c/li\u003e\n\u003cli\u003eShukri, A. K.; Mubarak, A. S. Factors of Academic Success Among Undergraduate Medical Students in Taif University, Saudi Arabia: A Cross-Sectional Study. 2019.\u003c/li\u003e\n\u003cli\u003eMonteverde-Su\u0026aacute;rez, D.; Gonz\u0026aacute;lez-Flores, P.; Santos-Sol\u0026oacute;rzano, R.; Garc\u0026iacute;a-Minjares, M.; Zavala-Sierra, I.; de la Luz, V. L.; S\u0026aacute;nchez-Mendiola, M. Predicting Students\u0026rsquo; Academic Progress and Related Attributes in First-Year Medical Students: An Analysis with Artificial Neural Networks and Na\u0026iuml;ve Bayes. BMC Medical Education 2024, 24 (1), 74. https://doi.org/10.1186/s12909-023-04918-6.\u003c/li\u003e\n\u003cli\u003eEsmat, T. A.; Pitts, J. D. Predicting Success in an Undergraduate Exercise Science Program Using Science-Based Admission Courses. Advances in Physiology Education 2020, 44 (2), 138\u0026ndash;144. https://doi.org/10.1152/advan.00130.2019.\u003c/li\u003e\n\u003cli\u003eHuberts, L. C. E.; Schoonhoven, Marit; and Does, R. J. M. M. Multilevel Process Monitoring: A Case Study to Predict Student Success or Failure. Journal of Quality Technology 2022, 54 (2), 127\u0026ndash;143. https://doi.org/10.1080/00224065.2020.1828008.\u003c/li\u003e\n\u003cli\u003eBir, D.; Ahn, B. Factors Predicting Students\u0026rsquo; Persistence and Academic Success in an Aerospace Engineering Program*.\u003c/li\u003e\n\u003cli\u003eO. Embarak. Apply Machine Learning Algorithms to Predict At-Risk Students to Admission Period. In 2020 Seventh International Conference on Information Technology Trends (ITT); 2020; pp 190\u0026ndash;195. https://doi.org/10.1109/ITT51279.2020.9320878.\u003c/li\u003e\n\u003cli\u003eZhang Feng; Chen Jingjing. A Review of Learning Data Characteristics Suitable for Student Performance Prediction. Software engineering, 2023, 26 (10), 1-4. https://doi.org/10.19644/j.cnki.issn2096-1472.2023.010.001.\u003c/li\u003e\n\u003cli\u003eMerchant, A. A. H.; Afzal, N.; Rahim, K. A.; Shah, S. A.; Jamal, W. Z.; Rahim, A.; Bakhshi, S. K.; Shaikh, N. Q.; Noorali, A. A.; Mahmood, S. B. Z.; Akbarali, L.; Tariq, M.; Khan, S.; Haider, A. H. Application to Achievement: Association between Pre-Admission Factors, Admission Scores, and Medical Students\u0026rsquo; Performance. BMC Medical Education 2025, 25 (1), 223. https://doi.org/10.1186/s12909-025-06800-z.\u003c/li\u003e\n\u003cli\u003eV. L. Uskov; J. P. Bakken; A. Byerly; A. Shah. Machine Learning-Based Predictive Analytics of Student Academic Performance in STEM Education. In 2019 IEEE Global Engineering Education Conference (EDUCON); 2019; pp 1370\u0026ndash;1376. https://doi.org/10.1109/EDUCON.2019.8725237.\u003c/li\u003e\n\u003cli\u003eFeng, J.; Shen, C.; Zhang, D.; Yang, W.; Xu, G. Development and Validation of a Nomogram to Predict Hemiplegic Shoulder Pain in Patients With Stroke: A Retrospective Cohort Study. Archives of Rehabilitation Research and Clinical Translation 2022, 4 (3), 100213. https://doi.org/10.1016/j.arrct.2022.100213.\u003c/li\u003e\n\u003cli\u003eKim, Y. H.; Jung, S. J.; Yang, E. J.; Paik, N. J. Clinical and Sonographic Risk Factors for Hemiplegic Shoulder Pain: A Longitudinal Observational Study. J Rehabil Med 2014, 46 (1), 81\u0026ndash;87. https://doi.org/10.2340/16501977-1238.\u003c/li\u003e\n\u003cli\u003eYue, S.; Li, S.; Huang, X.; Liu, J.; Hou, X.; Zhao, Y.; Niu, D.; Wang, Y.; Tan, W.; Wu, J. Machine Learning for the Prediction of Acute Kidney Injury in Patients with Sepsis. J Transl Med 2022, 20 (1), 215. https://doi.org/10.1186/s12967-022-03364-0.\u003c/li\u003e\n\u003cli\u003eWang, K.; Tian, J.; Zheng, C.; Yang, H.; Ren, J.; Liu, Y.; Han, Q.; Zhang, Y. Interpretable Prediction of 3-Year All-Cause Mortality in Patients with Heart Failure Caused by Coronary Heart Disease Based on Machine Learning and SHAP. Comput Biol Med 2021, 137, 104813. https://doi.org/10.1016/j.compbiomed.2021.104813.\u003c/li\u003e\n\u003cli\u003eShu, L.; Yan, H.; Wu, Y.; Yan, T.; Yang, L.; Zhang, S.; Chen, Z.; Liao, Q.; Yang, L.; Xiao, B.; Ye, M.; Lv, S.; Wu, M.; Zhu, X.; Hu, P. Explainable Machine Learning in Outcome Prediction of High-Grade Aneurysmal Subarachnoid Hemorrhage. Aging (Albany NY) 2024, 16 (5), 4654\u0026ndash;4669. https://doi.org/10.18632/aging.205621.\u003c/li\u003e\n\u003cli\u003eShen Jiahao, Guan Ying, Ou Chunquan, Tan Xiaoxue, Yu Shuang, Ling Zhihai, Chen Zheng. Analysis of Influencing Factors of Medical Master\u0026rsquo;s Degree Course Examination Scores Based on Structural Equation Model [J] Chinese Health Statistics,2022,39(05):695-698.\u003c/li\u003e\n\u003cli\u003eMa Chao, Huang Huaguo, Tian Xin, Liu Bingjie, Wen Kunjian, Wang Pengjie. Research on the Estimation Method of Single Wood Volume Based on Backpack LiDAR [J]. Remote Sensing Technology and Application, 2022, 37 (05): 1071-1083.\u003c/li\u003e\n\u003cli\u003eCui Tingting, An Xuelian, Sun Deliang, Chen Dongsheng, Zhu Youchen. Evaluation Model of landslide Susceptibility Based on SHAP interpretable Machine Learning [J]. Journal of Chengdu University of Technology (Natural Science Edition), 2025, 52 (01): 153-172. \u003c/li\u003e\n\u003cli\u003eDeBerard, M. S.; Spielmans, G. I.; Julka, D. L. Predictors of Academic Achievement and Retention among College Freshmen: A Longitudinal Study. College Student Journal 2004, 38 (1), 66\u0026ndash;80.\u003c/li\u003e\n\u003cli\u003eYe, Z.; An, S.; Gao, Y.; Xie, E.; Zhao, X.; Guo, Z.; Li, Y.; Shen, N.; Zheng, J. Association between the Triglyceride Glucose Index and In-Hospital and 1-Year Mortality in Patients with Chronic Kidney Disease and Coronary Artery Disease in the Intensive Care Unit. Cardiovasc Diabetol 2023, 22 (1), 110. https://doi.org/10.1186/s12933-023-01843-2.\u003c/li\u003e\n\u003cli\u003eFarzipour, A.; Elmi, R.; Nasiri, H. Detection of Monkeypox Cases Based on Symptoms Using XGBoost and Shapley Additive Explanations Methods. Diagnostics (Basel) 2023, 13 (14), 2391. https://doi.org/10.3390/diagnostics13142391.\u003c/li\u003e\n\u003cli\u003eShu, L.; Yan, H.; Wu, Y.; Yan, T.; Yang, L.; Zhang, S.; Chen, Z.; Liao, Q.; Yang, L.; Xiao, B.; Ye, M.; Lv, S.; Wu, M.; Zhu, X.; Hu, P. Explainable Machine Learning in Outcome Prediction of High-Grade Aneurysmal Subarachnoid Hemorrhage. Aging (Albany NY) 2024, 16 (5), 4654\u0026ndash;4669. https://doi.org/10.18632/aging.205621.\u003c/li\u003e\n\u003cli\u003eZhang Ze, Chu Zhe. Research on Student Performance Prediction and SHAP Feature Analysis Based on XGBoost Informationization Research. 2024; 50 (03): 34-40.\u003c/li\u003e\n\u003cli\u003eZan Ziyi, Yue Weifeng, Zhao Hangzheng, Cao Changming, Hu Jingdan, Hu Litang, Xu Yang, Chen Aiping. Construction of Agricultural Water Consumption Prediction Model Based on Machine Learning and SHAP Algorithm [J]. Water Saving Irrigation, 2024, (12): 102-110.\u003c/li\u003e\n\u003cli\u003eLievens, F.; Buyse, T.; Sackett, P. R. Retest Effects in Operational Selection Settings: Development and Test of a Framework. Personnel Psychology 2005, 58 (4), 981\u0026ndash;1007. https://doi.org/10.1111/j.1744-6570.2005.00713.x.\u003c/li\u003e\n\u003cli\u003eCuddy, M. M.; Swanson, D. B.; Clauser, B. E. A Multilevel Analysis of Examinee Gender and USMLE Step 1 Performance. Acad Med 2008, 83 (10 Suppl), S58-62. https://doi.org/10.1097/ACM.0b013e318183cd65.\u003c/li\u003e\n\u003cli\u003eCuddy, M. M.; Swanson, D. B.; Clauser, B. E. A Multilevel Analysis of the Relationships between Examinee Gender and United States Medical Licensing Exam (USMLE) Step 2 CK Content Area Performance. Acad Med 2007, 82 (10 Suppl), S89-93. https://doi.org/10.1097/ACM.0b013e3181400379.\u003c/li\u003e\n\u003cli\u003eQiao Lianquan, Li Lingling. The Influence of Parents\u0026rsquo; Educational Attainment on the Self-Positioning of Graduate Students: A Case Study of Doctoral Students at Xiamen University [J]. Hubei College Entrance Examination, 2011, (30): 7-11.\u003c/li\u003e\n\u003cli\u003eObregon-Cuesta, A. I.; M\u0026iacute;nguez-M\u0026iacute;nguez, L. A.; Le\u0026oacute;n-Del-Barco, B.; Mendo-L\u0026aacute;zaro, S.; Fern\u0026aacute;ndez-Solana, J.; Gonz\u0026aacute;lez-Bernal, J. J.; Gonz\u0026aacute;lez-Santos, J. Bullying in Adolescents: Differences between Gender and School Year and Relationship with Academic Performance. Int J Environ Res Public Health 2022, 19 (15), 9301. https://doi.org/10.3390/ijerph19159301.\u003c/li\u003e\n\u003cli\u003eYusuf, M.; Adigun, J. The Influence of School Sex, Location and Type on Students\u0026rsquo; Academic Performance. International Journal of Educational Sciences 2010, 2, 81\u0026ndash;85. https://doi.org/10.1080/09751122.2010.11889992.\u003c/li\u003e\n\u003cli\u003eTitus, A.; Adu, E. School Location and Gender as Correlates of Students\u0026amp;#39; Academic Achievement in Economics.\u003c/li\u003e\n\u003cli\u003eGoel, A. D.; Akarte, S. V.; Agrawal, S. P.; Yadav, V. Longitudinal Assessment of Depression, Stress, and Burnout in Medical Students. J Neurosci Rural Pract 2016, 7 (4), 493\u0026ndash;498. https://doi.org/10.4103/0976-3147.188625.\u003c/li\u003e\n\u003cli\u003eScott, P. (2020). Access to Postgraduate Study: Representation and Destinations. Scottish Government - Commissioner for Fair Access.\u003c/li\u003e\n\u003cli\u003eWakeling, P., \u0026amp; Laurison, D. (2017). Are Postgraduate Qualifications the \u0026lsquo;New Frontier of Social Mobility\u0026rsquo;? British Journal of Sociology,68(3), 533\u0026ndash;555.\u003c/li\u003e\n\u003cli\u003eNabizadeh, S.; Hajian, S.; Sheikhan, Z.; Rafiei, F. Prediction of Academic Achievement Based on Learning Strategies and Outcome Expectations among Medical Students. BMC Med Educ 2019, 19 (1), 99. https://doi.org/10.1186/s12909-019-1527-9.\u003c/li\u003e\n\u003cli\u003ePerrone, K.; Tschopp, M.; Snyder, E.; Boo, J.; Hyatt, C. A Longitudinal Examination of Career Expectations and Outcomes of Academically Talented Students 10- and 20-Years Post\u0026mdash;High School Graduation. Journal of Career Development - J CAREER DEVELOPMENT 2010, 36, 291\u0026ndash;309. https://doi.org/10.1177/0894845309359347.\u003c/li\u003e\n\u003cli\u003eKikuchi, A.; Kawamoto, R.; Abe, M.; Ninomiya, D.; Tokumoto, Y.; Kumagi, T. Ambiguous Motivations in Medical School Applicants: A Retrospective Study from Japan. Med Educ Online 2025, 30 (1), 2467487. https://doi.org/10.1080/10872981.2025.2467487.\u003c/li\u003e\n\u003cli\u003eWang Hui. Investigation and Research on the Admission Motivations of Master\u0026rsquo;s Degree Students with Interdisciplinary Background in Education [Master\u0026rsquo;s Thesis] Shanghai Normal University 2016.\u003c/li\u003e\n\u003cli\u003eZhu Xiaowen, Ding Wenjun, Zhang Mingliang. Research on the Current Situation and Enhancement Strategies of Positive Psychological Capital of Students in Police Colleges: Taking the Students of the 2017 Grade of Shandong Police College as Samples [J] Journal of Public Security (Journal of Zhejiang Police College), 2018, (06): 90-94.\u003c/li\u003e\n\u003cli\u003eMateos-Gonz\u0026aacute;lez, J. L.; Wakeling, P. Exploring Socioeconomic Inequalities and Access to Elite Postgraduate Education among English Graduates. Higher Education 2022, 83 (3), 673\u0026ndash;694. https://doi.org/10.1007/s10734-021-00693-9.\u003c/li\u003e\n\u003cli\u003eLiu, Y.; Hau, K.-T.; Zheng, X. Does Instrumental Motivation Help Students with Low Intrinsic Motivation? Comparison between Western andConfucian Students. International journal of psychology: Journal international de psychologie 2020, 55 (2), 182\u0026ndash;191. https://doi.org/10.1002/ijop.12563.\u003c/li\u003e\n\u003cli\u003eHyun, M. S. Examining the Validity of the GRE General Test Scores and Undergraduate GPA for Predicting Success in Graduate School at a Large Racially and Ethnically Diverse Public University in Southeast Florida. FIU Electronic Theses and Dissertations 2012. https://doi.org/10.25148/etd.FI12121002.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1. Description of postgraduate feature variables\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"659\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCoding scheme\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDemographic Characteristics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003egender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=female;2=male\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eprovince\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=Guangdong;2=non-Guangdong\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eonly_child\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003ehousehold_registration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=urban;2=rural\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003ehighest_degree_parents\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=illiterate;2=primary school;3=middle school;4=high school;5=bachelor degree;6=master/Ph.D\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003efather_occupation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=government office, national enterprise, and institution;2=worker;3=farmer;4=businessman;5=service practitioner;6=soldier;7=production and transportation personnel;8=other\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003emother_occupation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=government office, national enterprise, and institution;2=worker;3=farmer;4=businessman;5=service practitioner;6=soldier;7=production and transportation personnel;8=other\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eUndergraduate Academic Profile\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eundergraduate_GPA_rank\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=top 5%;2=top 10%;3=top 20-50%;4=below 50%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eundergraduate_colleges\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=985 university;2=211 university;3=SMU and others\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eundergraduate_score_average\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=60-69;2=70-79;3=80-89;4=\u0026ge;90\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eundergraduate_publication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=0;2=\u0026ge;1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003efirst_author_publication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eSCI_publication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eCET-4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=pass;2=fail to pass\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eCET-6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=pass;2=fail to pass\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eIELTS_or_TOEFL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=pass;2=fail to pass\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003ephysician\u0026rsquo;s_medical_license\u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no need\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003epunishment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003efrequency_activities\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=often;2=occasionally;3=never\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003enegative_behaviors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=often;2=occasionally;3=never\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003estudy_behaviors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=often;2=occasionally;3=never\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eself-improvement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=often;2=occasionally;3=never\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003erecreation_duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=\u0026le;7 hours;2=\u0026gt;7 hours\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003estudy_duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=\u0026le;7 hours;2=\u0026gt;7 hours\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePostgraduate Transition Factors\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003ereputation_SMU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=very important;2=somewhat important;3=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003ereasonable_tuition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=very important;2=somewhat important;3=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003esatisfied_major\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=very satisfied;2=somewhat satisfied;3=a little dissatisfied;4=very dissatisfied\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003etransfer_school\u003csup\u003ed\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003efresh_graduate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=yes;2=no\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003edegree_category\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=academic master;2=professional master\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003ehighest_degree_expectation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 451px;\"\u003e\n \u003cp\u003e1=none;2=master\u0026rsquo;s degree;3=doctor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003ehighest_degree_expectation_SMU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 451px;\"\u003e\n \u003cp\u003e1=none;2=master\u0026rsquo;s degree;3=doctor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003ecareer_expectation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 451px;\"\u003e\n \u003cp\u003e1=doctor;2=college teacher;3=scientific researchers;4=employee of the enterprise;5=administrator;6=other\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSelf-assessment metrics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003econtribute_to_country\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=absolutely necessary;2=very important;3=somewhat important;4=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eartistic_achievement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=absolutely necessary;2=very important;3=somewhat important;4=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003erecognized_in_work\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=absolutely necessary;2=very important;3=somewhat important;4=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003epursue_higher_degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=very important;2=somewhat important;3=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003emake_money\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=very important;2=somewhat important;3=insignificance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003econfidence\u003csup\u003ee\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eproblem-solving_ability\u003csup\u003ef\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003elearning_technical_ability\u003csup\u003eg\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=much higher;2=a little higher;3=equality;4=a little lower;5=much lower\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003econtinue_study\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 451px;\"\u003e\n \u003cp\u003e1=stand a good chance;2=have a chance;3=a very small chance;4=have no chance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 659px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e: It refers to the only child born to a couple. \u003cstrong\u003eb\u003c/strong\u003e: The calculation method of this variable is \u0026ldquo;weighted average score = (\u0026sum; [(a certain subject score * the same subject credit)])/total credits\u0026rdquo;. \u003cstrong\u003ec\u003c/strong\u003e: It refers to the qualification certificate uniformly issued by the National Health Commission after passing the national unified practicing physician qualification examination and practicing assistant physician qualification examination. It is a necessary certificate for practicing physicians in China. \u003cstrong\u003ed\u003c/strong\u003e: It means that during the postgraduate admission process, if a candidate meets the Re-examination conditions but cannot participate in the Re-examination at the first-choice institution or fails the Re-examination，the candidate can choose to transfer to other institutions. \u003cstrong\u003ee\u003c/strong\u003e: It refers to how an individual perceives their own level of self-confidence compared with the average level of their peers. \u003cstrong\u003ef\u003c/strong\u003e: It refers to how an individual perceives their own problem-solving ability compared with the average level of their peers. \u003cstrong\u003eg\u003c/strong\u003e: It refers to how an individual perceives their own learning technical ability compared with the average level of their peers.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2. Comparison of master\u0026rsquo;s student data between training set and test set\u003c/strong\u003e\u003c/p\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"658\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u003cstrong\u003etraining set\u003c/strong\u003e\u003cstrong\u003e（\u003c/strong\u003e\u003cstrong\u003en=655\u003c/strong\u003e\u003cstrong\u003e）\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003etesting set\u003c/strong\u003e\u003cstrong\u003e（\u003c/strong\u003e\u003cstrong\u003en=436\u003c/strong\u003e\u003cstrong\u003e）\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u003cimg width=\"15\" height=\"24\" src=\"https://myfiles.space/user_files/83062_751fab6dfaef2446/83062_custom_files/img1757586942.gif\" alt=\"image\"\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eDemographic characteristics\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003egender\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.168\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.280\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e242(36.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e168(39.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003efemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e425(63.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e257(60.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eprovince\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.217\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.642\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eGuangdong\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e273(40.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e180(42.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003enon-Guangdong\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e394(59.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e245(57.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eonly child\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e225(33.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e141(33.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.849\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ehousehold_registration\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.506\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.477\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eurban\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e349(52.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e213(50.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003erural\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e318(47.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e212(49.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ehighest_degree_parents\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e10.993\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.052\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eilliterate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e2(0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e0(0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eprimary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e52(7.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e51(12.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eMiddle school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e250(37.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e150(35.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ehigh school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e196(29.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e133(31.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ebachelor degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e142(21.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e84(19.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emaster/Ph.D\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e25(3.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e7(1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003efather_occupation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e10.652\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.155\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003egovernment/ national enterprise and institution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e105(15.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e60(14.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eworker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e99(14.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e85(20.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003efarmer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e155(23.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e97(22.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ebusinessman\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e79(11.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e53(12.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eservice practitioner\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e20(3.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e19(4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003esoldier\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e3(0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e0(0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eproduction and transportation personnel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e24(3.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e16(3.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eother\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e182(27.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e95(22.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003emother_occupation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e6.755\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.344\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003egovernment/national enterprise and institution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e102(15.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e50(11.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eworker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e77(11.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e64(15.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003efarmer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e155(23.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e102(24.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ebusinessman\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e46(6.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e33(7.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eservice practitioner\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30(4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e25(5.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003esoldier\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e0(0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e0(0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eproduction and transportation personnel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30(4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e25(5.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eother\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e250(37.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e147(34.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eCharacteristics of undergraduate period\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eundergraduate_GPA_rank\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.348\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.503\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003etop 5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e107(16.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e68(16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003etop 10%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e173(25.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e94(22.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003etop 20-50%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e318(47.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e213(50.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ebelow 50%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e69(10.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e50(11.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eundergraduate_colleges\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.414\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e985 university\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e11(1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e12(2.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e211 university\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e35(5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e23(5.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eSMU and others\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e621(93.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e390(91.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eundergraduate_score_average\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e6.833\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.077\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e60-69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e18(2.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e16(3.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e70-79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e239(35.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e165(38.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e80-89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e401(60.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e231(54.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u0026ge;90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e9(1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e13(3.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eundergraduate_publication \u0026ge;1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e107(16.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e66(15.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003efirst author publication paper\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e53(7.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e23(5.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e3.598\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e1.165\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003epublished SCI paper during undergraduate\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e29(4.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e18(4.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.052\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.974\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003epassed CET-4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e659(98.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e415(97.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.131\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.144\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003epassed CET-6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e530(79.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e328(77.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.370\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003epassed IELTS_or_TOEFL\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e35(5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e8(1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e7.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eacquired physician\u0026rsquo;s_medical_license\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e32(4.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e17(4.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.433\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.295\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eno punishment\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e665(99.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e425(100)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.277\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.259\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eoften frequency_activities\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e89(13.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e55(12.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.327\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.849\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eoften negative_behaviors\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e2(0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e0(0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.464\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.481\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eoften study_behaviors\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e286(42.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e190(44.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.448\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.485\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eoften self-improvement\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e142(21.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e92(21.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.918\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.632\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003erecreation duration\u0026gt;7 hours\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e9(1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e8(1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.497\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.869\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003estudy duration\u0026gt;7 hours\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e87(13.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e47(11.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e4.159\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eCharacteristics as a Postgraduate candidate\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSMU\u0026rsquo;s reputation is very important\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e326(48.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e221(52.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.437\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.487\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ereasonable SMU\u0026rsquo;s tuition is very important\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e187(28.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e146(34.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e4.888\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.087\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003esatisfied_major\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e644(96.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e409(96.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.897\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.826\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003efirst-choice candidate\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e451(67.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e286(67.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.912\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003efresh graduate\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e492(73.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e312(73.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.898\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003edegree_category\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.152\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.696\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eacademic master\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e492(73.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e318(74.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eprofessional master\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e175(26.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e107(25.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ehighest_degree_expectation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.326\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003enone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e36(5.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e17(4.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emaster\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e138(20.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e78(18.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003edoctor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e493(73.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e330(77.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ehighest_degree_expectation_SMU\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.708\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003enone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e22(3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e18(4.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emaster\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e253(37.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e157(36.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003edoctor\u0026rsquo;s degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e392(58.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e250(58.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ecareer_expectation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e5.484\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.360\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003edoctor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e263(39.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e161(37.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ecollege teacher\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e108(16.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e70(16.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003escientific researchers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e137(20.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e99(23.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eemployee of the enterprise\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e35(5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e30(7.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eadministrator\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e47(7.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e19(4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eother\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e77(11.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e46(10.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eSelf-cognitive assessments.\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emore confident than others\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e206(30.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e144(33.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.446\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.836\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ea better problem solver than others\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e184(27.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e148(34.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e8.342\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.039\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ea better learner of technology than others\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e113(16.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e80(34.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e1.184\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.881\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003ea good chance to further study\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e286(42.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e203(47.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e5.569\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.135\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eabsolutely necessary to make a contribution to the country\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e71(10.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e61(14.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e6.640\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eabsolutely necessary to make achievements in art\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e22(3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e23(5.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e7.940\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.047\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eabsolutely necessary to be recognized in your field\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e172(25.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e116(27.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e3.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.275\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003eget a higher degree is very important\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e463(69.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e299(70.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.824\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.662\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 341px;\"\u003e\n \u003cp\u003emake money is very important\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 95px;\"\u003e\n \u003cp\u003e508(76.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 107px;\"\u003e\n \u003cp\u003e329(77.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 62px;\"\u003e\n \u003cp\u003e2.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.271\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3. Predictive performance of the eight machine learning techniques\u003c/strong\u003e\u003c/p\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"615\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003eRecall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003eF1-score\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eFDA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.705\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.825\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.966\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.901\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eLDA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.724\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.816\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.851\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.941\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.894\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eMDA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.710\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.791\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.845\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.878\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.735\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.823\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.838\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.972\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.900\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.848\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.969\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.905\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.818\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.823\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.992\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.900\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.666\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.823\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.901\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 87px;\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.744\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 115px;\"\u003e\n \u003cp\u003e0.828\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.969\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.902\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" style=\"width: 615px;\"\u003e\n \u003cp\u003eNote: AUC, the value of the area under the three classification curves; FDA, Flexible Discriminant Analysis; LDA, Linear Discriminant Analysis; MDA, Mixture Discriminant Analysis; GBM, Gradient Boosting Machine; LR, Logistic Regression; RF, Random Forest; SVM, Support Vector Machine; XGBoost, Extreme Gradient Boosting.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-education","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"meed","sideBox":"Learn more about [BMC Medical Education](http://bmcmededuc.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/meed/default.aspx","title":"BMC Medical Education","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"GPA prediction, Boruta algorithm, SHAP model, machine learning, XGBoost model","lastPublishedDoi":"10.21203/rs.3.rs-7186245/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7186245/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eIn recent years, Chinese medical postgraduate education has undergone significant transformation, with enrollment soaring to 156,000 students in 2023, accounting for 12% of the nation\u0026rsquo;s total postgraduate admissions. Recognizing high-achieving students at an early stage and learning from their success can change the way to influence future educational dilemmas. While the existing evaluation system remained limited in its ability to prospectively predict academic performance variability. This critical gap underscores the need for innovative, data-driven approaches to transcend conventional assessment paradigms. Our study used machine learning techniques to predict the academic outcomes (GPA) of medical postgraduate students in a certain university, providing evidence-based strategies that could be used to improve educational practices and improve student performance.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eIn this study, we worked with 1,133 postgraduate students at Southern Medical University (2020 cohort) while analyzing 42 variables, including demographic, undergraduate performance, postgraduate transition variables, and measures of their self-assessment. Using the Boruta algorithm, we identified the most important predicting features and then tested eight learning machine models to find the best one. Furthermore, we applied SHAP (Shapley Additive Explanations) to derive interpretable insights into the most critical features of success. Finally, we submitted this work as an interactive web application that allowed academic leaders to predict their students\u0026rsquo; GPAs and provide much-needed proactive support.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eXGBoost model crushed the competitions, delivering higher predictions (AUC\u0026thinsp;=\u0026thinsp;0.744, Accuracy\u0026thinsp;=\u0026thinsp;82.8%, F1\u0026thinsp;=\u0026thinsp;0.902). SHAP analysis exposed the secret formula for success. A student\u0026rsquo;s career ambition, undergraduate ranking, and core subject mastery weren\u0026rsquo;t just important, but they were also game changers. Based on the results of our study, we built a smart web tool that could turn data into action, giving educators a crystal ball to flag high-potential students early, personalize support proactively, and allocate resources smartly. Our results were not just number crunching; it will also be a new playbook for medical postgraduate students\u0026rsquo; education.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eThis research showed that XGBoost model did not just predict academic performance; it also revealed hidden pathways for students\u0026rsquo; success. We found that students should focus on ambition, track record, and their mastery of core subjects to drive their success based on machine learning and SHAP models. More importantly, we developed a clever, user-friendly tool that might help educators identify students with high potential earlier and intervene with support ahead of challenges.\u003c/p\u003e","manuscriptTitle":"Enhancing Graduate Education Assessment: A Machine Learning Approach to GPA Prediction for Medical Students","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-11 10:38:10","doi":"10.21203/rs.3.rs-7186245/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-03T07:05:57+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-24T07:59:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"284764928476893173157935282011439662930","date":"2025-09-24T06:12:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"306950268405140186711811128850183306835","date":"2025-09-22T04:54:44+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-19T12:08:20+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"113842777648084500910738805766789727527","date":"2025-09-09T08:37:10+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"269012496236880686795530008040076922654","date":"2025-09-06T17:21:34+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-04T15:16:09+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-03T03:26:44+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-18T10:07:52+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-16T06:52:45+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Education","date":"2025-08-16T06:49:12+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-education","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"meed","sideBox":"Learn more about [BMC Medical Education](http://bmcmededuc.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/meed/default.aspx","title":"BMC Medical Education","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f0d0ac89-2919-4042-82be-3a0048de20e4","owner":[],"postedDate":"September 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-02-16T16:05:17+00:00","versionOfRecord":{"articleIdentity":"rs-7186245","link":"https://doi.org/10.1186/s12909-026-08741-7","journal":{"identity":"bmc-medical-education","isVorOnly":false,"title":"BMC Medical Education"},"publishedOn":"2026-02-10 15:58:12","publishedOnDateReadable":"February 10th, 2026"},"versionCreatedAt":"2025-09-11 10:38:10","video":"","vorDoi":"10.1186/s12909-026-08741-7","vorDoiUrl":"https://doi.org/10.1186/s12909-026-08741-7","workflowStages":[]},"version":"v1","identity":"rs-7186245","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7186245","identity":"rs-7186245","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}