Artificial Intelligence for Predicting Arabica Coffee Yield Based on Climate and Soil in the State of Minas Gerais, Brazil

Artificial Intelligence for Predicting Arabica Coffee Yield Based on Climate and Soil in the State of Minas Gerais, Brazil | 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 Artificial Intelligence for Predicting Arabica Coffee Yield Based on Climate and Soil in the State of Minas Gerais, Brazil Rogério William Fernandes Barroso¹, Adriano Bortolotti Silva², and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8064468/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Coffee is an agricultural commodity of great economic importance for Brazil and worldwide. In this context, the country also stands out as the largest producer and exporter of Arabica coffee. Furthermore, coffee consumption in Brazil grew by 1.64%, reflecting its social and economic relevance. The state of Minas Gerais leads national production, with more than 1 million hectares under cultivation and accounting for over 70% of the gross production value of Arabica coffee in 2023. Factors such as climate and soil directly contribute to this result, impacting yield. This research aimed to study temperature variations and optimal soil conditions so that farmers can plan and forecast yield expectations year after year. In this context, this study applied predictive artificial intelligence models, such as TD, MLP, RF, XGBoost, SVM, GBR, and KNN, including in the analyses variables such as climate, soil, and coffee yield, using data from 200 municipalities in Minas Gerais. The goal was to support farmers in estimating their production, contributing to better harvest planning and optimizing production costs. Artificial intelligence Yield Predictive models Arabica coffee Figures Figure 1 Figure 2 Figure 3 Figure 4 1. INTRODUCTION Coffee is an agricultural commodity that generates wealth and holds an important position in both national and global economies. Brazil stands out as the world’s largest producer and exporter of Arabica coffee, accounting for 42% of global production. 1 According to ABIC (2023), coffee is present in Brazilian households, showing a 1.64% increase in consumption in 2023 compared to the same period of the previous year. 2 This increase in consumption reflects the importance of coffee farming in the economic and social development of the country, significantly contributing to income generation. 3 The states of Minas Gerais, São Paulo, and Espírito Santo stand out for their extensive production areas, totaling 1,422,715 hectares in 2024, according to EMBRAPA (2024). In particular, the state of Minas Gerais stands out as the largest coffee-producing state in Brazil, with 1,108,211 hectares. 1 In 2023, the gross production value of Arabica coffee in Brazil was R $ 36,804,111.19, with the Southeast region accounting for R $ 34,274,655.55 and the state of Minas Gerais reaching R $ 26,539,536.46, representing more than 70% of the gross production value of the entire country, according to a report by MAPA (2023). 4 These data are consistent with the favorable conditions for coffee cultivation in Brazil. ACEVES-NAVARRO et al. (2020) reported that the optimal temperatures for coffee production range from 19 to 22°C, which can be found in the state of Minas Gerais, characterized by an ideal combination of altitude, temperature, and precipitation, providing the necessary conditions for Arabica coffee cultivation (DAMATTA et al., 2018; DOORENBOS & KASSAM, 1979; TAVARES et al., 2018; LORENÇONE et al., 2021). Climatic variability directly influences the plant, resulting in fluctuations in yield, as evidenced by Läderach et al. (2017) and Monteiro et al. (2017). 5 – 11 The research also considered soil composition, a factor that directly impacts coffee production. According to Kouadio et al. (2018), soil pH, organic matter and nutrient content, texture, and structure can significantly influence fertility, and the absence of any of these aspects may result in less fertile soils, reduced water retention and aeration, decreased infiltration, increased erosion, and hindered supply of essential nutrients to plants. 12 A wide range of factors can affect yield, and different tools can support the understanding of results. These tools are artificial intelligence models, which represent an accessible alternative, mathematically less complex, facilitating process automation in computerized systems. Their application has been widely used in the agronomic field for yield forecasting of crops such as coffee and soybean, proving to be an ally in planning and decision-making at harvest time by farmers, enabling a deeper analysis of large datasets and variables that may affect coffee yield more simply. 9 , 13 – 15 Considering that studies have not used temporal data on climate, soil, and yield in machine learning models, this research aimed to investigate the application of Temporal Difference (TD), Multilayer Perceptron (MLP), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), Gradient Boosting Regressor (GBR), and K-Nearest Neighbors (KNN) models to understand the relationships among soil properties, climate variables, and Arabica coffee yield data obtained from 200 municipalities in the state of Minas Gerais. 2. MATERIAL AND METHODS 2.1. Study area The research was conducted in the state of Minas Gerais, using historical data series over 10 years (2011–2021) from 200 municipalities with high Arabica coffee production, according to IBGE (2023). 16 These municipalities were divided into twelve mesoregions for individualized analysis of climate and soil variables. Climate data were obtained from the National Aeronautics and Space Administration/Prediction of Worldwide Energy Resources (NASA/POWER) platform, a tool validated in several regions of the world through studies comparing surface data, demonstrating its accuracy with respect to maximum, minimum, and mean temperature parameters in Egyptian territories. 17 , 18 The Application Programming Interface (API) ( https://power.larc.nasa.gov/api/pages/ ) was used as the tool to retrieve the following variables: air temperature at 2 meters (T2M), maximum (T2M_MAX) and minimum (T2M_MIN) temperature at the same height, relative humidity at 2 meters (RH2M), precipitation (PRECTOTCORR), surface pressure (PS), and wind speed at 10 meters (WS10M). Soil data were retrieved from the SoilGrids platform, a global information system containing spatial predictions for various soil properties (TIFAFI et al., 2017), through the SoilGrids REST API 2.0 ( https://rest.isric.org/soilgrids/v2.0/docs ). The following input variables were considered for this study: clay content (Clay), total nitrogen (Nitrogen), organic carbon density (OCD), organic carbon stocks (OCS), soil pH (PHH2O), and soil organic carbon content in the fine earth fraction (SOC). The data were organized by mesoregions of the state of Minas Gerais. The state of Minas Gerais has a wide variety of soils and microclimates, with coffee cultivation being practiced across all of them, particularly in the South/Southwest, Zona da Mata, and Triângulo Mineiro mesoregions, which are responsible for most of the coffee production. The IBGE Automatic Recovery System – SIDRA allowed us to perform the geographical analysis of the producing mesoregions. 2.2. Artificial intelligence models The use of artificial intelligence for yield prediction has been widely applied, especially in the field of agronomic sciences, for example, coffee yield forecasting, soybean yield forecasting through machine learning, soybean yield prediction, and sugarcane yield estimation. 9 , 13 , 15 , 21 However, no literature is found when it comes to comparing several predictive models for estimating Arabica coffee yield in various municipalities of Minas Gerais. According to James et al. (2013), there are several models aimed at classifying based on probabilities, one of which is the k-nearest neighbors (KNN) method. Given a positive integer K and a test observation x₀, the KNN classifier identifies the K points closest to x₀ in the training set, represented by N₀, and then estimates the conditional probability for class j as the fraction of points in N₀ whose response values are equal to j: 22 According to G. Biau & E. Scornet (2016), the Random Forest (RF) method can be described as M regression trees randomly distributed, creating several decision trees. This method works on the simple and effective logic of “divide and conquer,” splitting the data into smaller parts, building a random tree model for each segment, and subsequently integrating these models to form a consolidated prediction. This supervised algorithm can be applied to various types of problems, whether with small or large datasets, requiring only a small number of parameters. 23 Another widely used classification model is the Support Vector Machine (SVM). According to James et al. (2013), SVM gained great popularity in the 1990s due to its strong performance and distinct approach of finding a hyperplane that separates the data in the best possible way—something not present in classical approaches such as logistic regression and linear discriminant analysis. 22 The Multilayer Perceptron (MLP) is used in complex scenarios that require supervised learning algorithms to employ multiple hyperplanes and address problems that are not linearly separable. This model allows for multiple input and output layers, in addition to hidden layers that determine the number of neurons, which in turn define the model’s hyperplanes, enabling more flexible and nonlinear separations. In contrast, SVM applies a single optimal hyperplane to separate the classes. Temporal Difference (TD) is a concept of reinforcement learning. This technique consists of updating state values based on the difference between the current prediction and the observed future reward. According to Sutton and Barto (1998), this method addresses prediction problems using experiences accumulated over time. 24 The algorithm generates an immediate target at time step t + 1 and performs an update using the observed reward R t+1 and the estimate V( S t+1 ) of the next state. The simplest TD method performs this update without considering the value of states beyond the next one, adjusting directly after each transition, known as TD(0) : Extreme Gradient Boosting (XGBoost) is an efficient and widely used algorithm based on the boosting method. According to Alshboul et al. (2022), it was developed as an enhancement of the gradient boosting method, being mainly applied to regression and classification tree problems. 25 The concept of boosting forms its foundation, combining the prediction of weak learners with additive training methods to build a more robust and accurate model. Mathematically, the objective function of XGBoost is expressed as: where \(\:l\left({y}_{i},{y}_{i}^\right)\) represents the loss function between the prediction and the true value, and \(\:\varOmega\:\left({f}_{k}\right)\) is a regularization term that penalizes model complexity to avoid overfitting by capturing noise and irrelevant patterns, which would impair performance on new data. Finally, the Gradient Boosting Regressor (GBR), similar to XGBoost, is an ensemble learning model based on the boosting method. According to Otchere et al. (2021), GBR consists of an iterative collection of decision trees organized sequentially, where each tree learns from the errors of the previous one to improve model accuracy. Mathematically, a GBR with M trees can be represented as: where \(\:hm\left(xj\right)\) represents each weak learner tree, \(\:{\gamma\:}_{m}\) is the weight assigned to each tree, and M is the total number of trees. This process allows GBR to progressively reduce residual error, making it a powerful method for regression and classification problems. 2.3. Model evaluation metrics Regression analysis was performed to understand the relationships among soil properties, climate variables, and Arabica coffee yield data obtained from 200 municipalities in the state of Minas Gerais and to verify which model achieved the highest accuracy. The difference between the actual data points and the best-fit line produced by the algorithm is considered the model error. The model error with multiple data points was determined using the following criteria, characterized by JAMES et al. (2013): coefficient of determination (R 2 ), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Squared Error (MSE). These metrics are essential to ensure proper model evaluation and were applied in this study to select the best machine learning algorithm. 22 3. RESULTS The predictive models were subjected to several parameters to identify which achieved the best results among them (Table 1 ). The performance analysis demonstrates the accuracy of each machine learning model for predicting Arabica coffee yield in Minas Gerais and highlights the specific adequacy of each technique to the characteristics of the dataset used. Table 1 – Parameters defined and tested in each machine learning algorithm used to predict Arabica coffee yield in Minas Gerais. Parameter Tested Values TD max_depth min_samples_split min_samples_leaf random_state criterion 4, 5, 6, 7, 8, 9, 10 2,5,10,15,20 1,5,15,20 10 friedman_mse, poisson, squared_error, absolute_error MLP hidden layer sizes (9; 10; 9); (7; 8; 10); (10; 10; 10) max iter 100; 1000; 1500; 7000 cross-Validation 3 learning rate init 0.01; 5; 10 learning rate constant; adaptive Activation relu; tanh; logistic Solver adam, sgd Alpha 0.01; 5; 10 random state 20 SVM Kernel rbf; linear; poly; sigmoid C 0.1; 50; 100; 1000; 10000 Épsilon 0.1; 1; 4; 5; 10; 100 Gamma auto; scale Degree 1; 2; 3; 4; 7 coef0 0; 1 RF max depth none, 5, 10, 15 n estimators 10; 200; 500; 1000; 10000 random state 29 XGBOOST max depth 1;3;5; 10; 100 Alpha 0.1, 20; 200; 2000 n estimators learning_rate objective 100; 200;300; 3000 0.1, 0.01, 0.05 reg:squarederror GBR learning_rate n estimators 0.01, 0.05, 0.1 10; 100; 500; 1000 max depth values none; 5; 10; 100 random state values 29 KNN n neighbors 3;5;7;9;11 Weights Metric uniform, distance euclidean, Manhattan The models tested include RF, TD, SVM, KNN, GBR, XGBoost, and MLP, each with an optimal hyperparameter configuration adjusted to maximize prediction accuracy. For this purpose, the scikit-learn library method was employed. Table 2 – Application of the optimal parameters in the models and their results. Model Training Testing Model performance measures MAPE R² RMSE MSE MAE MAPE R² RMSE MSE MAE RF 0.057 0.999 5.87 34.48 1.04 0.177 0.995 65.48 4287.85 5.50 TD 0.105 0.999 4.26 18.13 1.61 0.209 0.996 55.68 3100.26 5.44 SVM 0.990 0.999 0.05 0.00 0.04 1.126 0.999 0.05 0.00 0.04 KNN 0.0004 0.999 0.00 0.00 0.00 2094.086 0.764 467.87 218905.09 276.64 GBR 2.487 0.999 1.62 2.61 0.70 3.214 0.996 55.00 3025.46 3.84 XGBOOST 25.877 0.999 7.34 53.83 3.03 152.569 0.996 60.91 3710.21 8.61 MLP 82.385 0.999 14.52 210.86 10.63 214.021 0.999 21.14 446.99 12.55 The RF model showed excellent accuracy in both training and testing, with R 2 close to 1 in both cases, indicating a high capacity of the model to explain the variance in the data. This model also presented low error values (MAE and RMSE) in both training and testing, standing out as one of the most robust models for this type of data. The TD model also showed high accuracy, with performance similar to RF during training, but with a slight increase in error values during testing, which may indicate the onset of overfitting. Nevertheless, it remains a solid choice due to its interpretability and ease of implementation. The SVM, known for its effectiveness in high-dimensional spaces, achieved virtually zero errors in training, reflecting an almost perfect adaptation to the training data, and maintained this characteristic in the test set. This performance suggests that the linear SVM model is extremely efficient for this specific dataset, possibly due to the linearity between the explanatory variables and the response. In contrast, the KNN model, despite showing perfect performance in training, exhibited a large discrepancy in testing, with extremely high errors, indicating a significant generalization issue for new data, perhaps due to its sensitivity to outliers or the distribution of the test data. Both GBR and XGBoost, as boosting models, achieved good results, with low error metrics during training and acceptable values in testing, demonstrating good generalization ability and robustness. These models are known for their efficiency in complex and heterogeneous datasets, which is typical of agricultural data. Lastly, the MLP, a type of neural network, showed the greatest disparity between training and testing results, indicating considerable overfitting despite its high training accuracy. SVM achieved the most accurate prediction of Arabica coffee yield among all the models analyzed. This result corroborates the findings of Kouadio et al. (2018), Kumar et al. (2018), and Kilic et al. (2024), who applied the model in agricultural contexts, providing farmers with robust and practical tools for forecasting yields, diseases, and pest infestations, integrating them into farm management systems, optimizing resource use, and maximizing production. 12 , 27 , 28 The same dataset used to predict coffee yield in the municipality of Muzambinho in 2021 was employed to perform tests with the trained model, where the average production yield was 1,200 kilograms per hectare, according to IBGE – SIDRA (2023). The prediction was 1,386 kilograms per hectare when submitted to the trained algorithm, a difference of 186 kilograms per hectare for the municipality. 29 4. DISCUSSION It is necessary to evaluate a range of factors related to production, given the influence and impact each one has on the results. Some studies address the yield associated with climate [1] and the biophysical aspects of the soil [2] separately. Therefore, we considered both aspects related to production in this study. Climate characteristics, such as temperature, affect coffee production. According to Lorençone et al. (2021), the ideal temperatures for coffee cultivation range from 19 to 22 °C, while leaf yellowing is observed above 25 °C, favoring the emergence of pests. Figures 2 and 3 show that all mesoregions of the state of Minas Gerais present suitable temperatures. However, rainfall conditions are not homogeneous. According to SILVIA & SOUZA (2022), yield was greater in regions with higher rainfall rates, as observed in Campo das Vertentes, Metropolitan Region of Belo Horizonte, West of Minas Gerais, South/Southeast of Minas Gerais, Triângulo Mineiro/ Alto Paranaíba, and Zona da Mata. 9,30 Soil plays an important role in the coffee production process, and its fertility is a highly dynamic process influenced by several factors, including nutrient availability, such as pH, organic matter, soil texture, and nitrogen, among others. According to Kouadio et al. (2018), soils with low pH are less favorable for nutrient absorption and are consequently less fertile. This pattern is standardized across all regions (Figure 4). Organic matter, which is important for increasing water retention, soil infiltration, aeration, and reducing erosion, in addition to supplying nutrients to plants, shows slight discrepancies among the regions. Ultisols, characterized by higher clay content across all mesoregions, predominate in the state of Minas Gerais. 12 Climate influences coffee yield, which can be confirmed in Graph 1, presenting the average production yield by mesoregion. According to Simões et al. (2015), the south of Minas Gerais is the rainiest, with an average annual precipitation of around 1,600 mm. This aligns with Graph 1, which exhibits that the South/Southwest of Minas Gerais is the region with the largest number of municipalities included in the analysis, showing a good distribution of production, even with some municipalities above and below the general average, followed by the Triângulo Mineiro/Alto Parnaíba, which also includes highly productive municipalities. 31 The graph shows large differences in yield among the regions of Minas Gerais. Some regions, such as the South/Southwest of Minas Gerais and Triângulo Mineiro, stood out for their higher yield and consistency, while others, such as Vale do Rio Doce and Vale do Mucuri, exhibited characteristics that were not favorable for coffee cultivation. Variations in climate can be observed, although soil properties do not vary substantially, as illustrated in the histograms. [1] LORENÇONE, João Antonio et al , PREVISÃO DA PRODUTIVIDADE DO CAFÉ COM BASE EM DADOS AGROCLIMÁTICOS E APRENDIZAGEM DE MÁQUINA / FORECASTING COFFEE YIELD BASED ON AGROCLIMATIC DATA AND MACHINE LEARNING, International Journal of Environmental Resilience Research and Science , v. 3, n. 1, 2021. [2] KOUADIO, Louis et al , Artificial intelligence approach for the prediction of Robusta coffee yield using soil fertility properties, Computers and Electronics in Agriculture , v. 155, p. 324–338, 2018. 5. CONCLUSION This study conducted a comparative analysis of several machine learning models, including SVM, Random Forest, Decision Tree, KNN, GBR, XGBoost, and MLP, to predict Arabica coffee yield in Minas Gerais, using six soil datasets (clay, total nitrogen, carbon density, carbon stock, pH, and carbon content) and seven climate datasets (2-meter air temperature, maximum and minimum 2-meter temperature, 2-meter humidity, rainfall, surface pressure, and 10-meter wind speed) as predictor variables, and coffee yield as the target variable. SVM proved to be the most efficient tested model when incorporating soil and climate data. The potential of integrating artificial intelligence algorithms with different crops is reinforced with this study, supporting farmers’ decision-making and improving harvest planning, as well as contributing to public policies aimed at coffee producers. Declarations Author Contribution R.W.F.B. conceived and conducted the research, performed data collection, and wrote the main manuscript text.A.B.S. supervised the study and contributed to the interpretation of the results and critical review of the manuscript.L.E.O. provided methodological guidance, assisted in data validation, and reviewed the final version of the manuscript.All authors reviewed and approved the final manuscript.The Corresponding Author, R.W.F.B., was responsible for all communications with the journal and ensuring compliance with Springer’s authorship policies. Acknowledgement I thank the Coordination for the Improvement of Higher Education Personnel (CAPES) for the financial support, the Federal Institute of Education, Science, and Technology of Southern Minas Gerais – Muzambinho Campus for the support and encouragement, and José do Rosário Vellano University (UNIFENAS) for the opportunity to enroll in the Doctoral Program. Finally, and no less importantly, I thank my advisors, Prof. Dr. Lucas Eduardo de Oliveira and Prof. Dr. Adriano Bortolotti da Silva, for their constant support, dedication, and guidance throughout this journey. References Embrapa (2024), Sumário executivo: Produção mundial de café estimada para safra 2023–2024 totaliza 171,4 milhões de sacas de 60 kg, Embrapa, publicado em 18 abril 2024, disponível em: https://www.embrapa.br/busca-de-noticias/-/noticia/88547345/producao-mundial-de-cafe-estimada-para-safra-2023-2024-totaliza-1714-milhoes-de-sacas-de-60kg (acesso em: 23 ago. 2024). Indicadores da Indústria de Café (2023), ABIC. Disponível em: https://estatisticas.abic.com.br/estatisticas/indicadores-da-industria/indicadores-da-industria-de-cafe-2023/. Acesso em: 17 mai. 2024. [Internet]. ABIC. [citado 16 de maio de 2024]. Disponível em: https://estatisticas.abic.com.br/estatisticas/indicadores-da-industria/indicadores-da-industria-de-cafe-2023/ Fassio, L. H. and Silva, A. E. S. (2015), ‘Importância econômica e social do café Conilon’, in: Livro Café Conilon 2007, Instituto Capixaba de Pesquisa, Assistência Técnica e Extensão Rural (Incaper), Espírito Santo, pp. 34–49. Em. VBP_10_23.pdf [Internet]. [citado 16 de maio de 2024]. Disponível em: http://www.consorciopesquisacafe.com.br/images/stories/noticias/2021/2023/outubro/VBP_10_23.pdf Aceves-Navarro LA, et al. Impact of climatic change on the adaptation of coffee (Coffea arabica L.) crops in Tabasco, Mexico. Agroproductividad. 2020;13(4):53–58. Damatta FM, et al. Physiological and agronomic performance of the coffee crop in the context of climate change and global warming: a review. J Agric Food Chem. 2018;66(21):5264–5274. DOORENBOS, J.; KASSAM, A. H. Yield response to water. [s.l: s.n.]. Tavares PDS, et al. Climate change impact on the potential yield of Arabica coffee in southeast Brazil. Reg Environ Change. 2018;18(3):873–83. Lorençone JA, Aparecido LEDO, Lorençone PA, Moraes JRDSCD. PREVISÃO DA PRODUTIVIDADE DO CAFÉ COM BASE EM DADOS AGROCLIMÁTICOS E APRENDIZAGEM DE MÁQUINA / FORECASTING COFFEE YIELD BASED ON AGROCLIMATIC DATA AND MACHINE LEARNING. Intern Journ Env Res Res Sci [Internet]. 5 de maio de 2021 [citado 16 de maio de 2024];3(1). Disponível em: https://e-revista.unioeste.br/index.php/ijerrs/article/view/26255 Läderach P, Ramirez-Villegas J, Navarro-Racines C, et al. Climate change adaptation of coffee production in space and time. Clim Change. 2017;141:47–62. doi:10.1007/s10584-016-1788-9. Monteiro JEB de A, et al. Modeling of corn yield in Brazil as a function of meteorological conditions and technological level. Pesq Agropec Bras. 2017;52(3):137–48. Kouadio L, Deo RC, Byrareddy V, Adamowski JF, Mushtaq S, Phuong Nguyen V. Artificial intelligence approach for the prediction of Robusta coffee yield using soil fertility properties. Computers and Electronics in Agriculture. dezembro de 2018;155:324–38. Torsoni GB, De Oliveira Aparecido LE, Dos Santos GM, Chiquitto AG, Da Silva Cabral Moraes JR, De Souza Rolim G. Soybean yield prediction by machine learning and climate. Theor Appl Climatol. fevereiro de 2023;151(3–4):1709–25. De Oliveira Aparecido LE, De Souza Rolim G, Camargo Lamparelli RA, De Souza PS, Dos Santos ER. Agrometeorological Models for Forecasting Coffee Yield. Agronomy Journal. janeiro de 2017;109(1):249–58. Guimarães EDS. Aprendizado de Máquina aplicado à predição da produtividade da cultura da soja utilizando dados de clima e solo [Internet] [Mestrado em Matemática, Estatística e Computação]. [São Carlos]: Universidade de São Paulo; 2020 [citado 16 de maio de 2024]. Disponível em: https://www.teses.usp.br/teses/disponiveis/55/55137/tde-09062020-123106/ IBGE (2023), Sistema IBGE de Recuperação Automática - SIDRA: produção agrícola municipal: tabelas. Minas Gerais. Disponível em: https://sidra.ibge.gov.br/tabela/5457. Acesso em: 5 mar. 2023. Stackhouse PW, Westberg D, Hoell JM, Chandler WS, Zhang T. Prediction of worldwide energy resource (POWER) – agroclimatology methodology (1.0 latitude by 1.0 longitude spatial resolution). NASA Tech Rep. 2015. Aboelkhair, H., Morsy, M. and Afandi, G. E. (2019), ‘Assessment of agroclimatology NASA POWER reanalysis datasets for temperature types and relative humidity at 2 m against ground observations over Egypt’, Advances in Space Research, Vol. 64, No. 1, pp. 129–42. Tifafi M, Guenet B, Hatté C. Large Differences in Global and Regional Total Soil Carbon Stock Estimates Based on SoilGrids, HWSD, and NCSCD: Intercomparison and Evaluation Based on Field Data From USA, England, Wales, and France. Global Biogeochemical Cycles. 2018;32(1):42–56. Simões, J. C. and Pelegrini, D. F. (2010), Diagnóstico da Cafeicultura Mineira – Regiões Tradicionais: Sul/Sudoeste de Minas, Zona da Mata, Triângulo Mineiro/Alto Paranaíba. EPAMIG, Belo Horizonte. (EPAMIG. Série Documentos, n. 46). Barbosa LAF, Pedronette DCG. ESTUDO COMPARATIVO ENTRE DIFERENTES REGRESSORES PARA ESTIMAR PRODUTIVIDADE DE CANA-DE-AÇÚCAR. James, G., Witten, D., Hastie, T. and Tibshirani, R. (2013), An Introduction to Statistical Learning: With Applications in R. Springer, New York. Biau G, Scornet E. A random forest guided tour. TEST. 1 o de junho de 2016;25(2):197–227. Sutton, R. S. and Barto, A. G. (1998), Reinforcement Learning. MIT Press, Cambridge, MA. Alshboul, O., Shehadeh, A., Almasabha, G. and Almuflih, A. S. (2022), ‘Extreme Gradient Boosting-Based Machine Learning Approach for Green Building Cost Prediction’, Sustainability, Vol. 14, No. 11, 6651. https://doi.org/10.3390/su14116651. Application of gradient boosting regression model for the evaluation of feature selection techniques in improving reservoir characterisation predictions. Kumar S, Mishra S, Khanna P, Pragya. Precision Sugarcane Monitoring Using SVM Classifier. 2018 [citado 9 de março de 2025]; Disponível em: https://arxiv.org/abs/1803.09413 Kilic I, Yaman O. Classification Method of Plants with Support Vector Machine (SVM) Using Local Binary Gaussian Model (LBGP) for Smart Agriculture. Em: 2024 32nd Signal Processing and Communications Applications Conference (SIU) [Internet]. Mersin, Turkiye: IEEE; 2024 [citado 9 de março de 2025]. p. 1–4. Disponível em: https://ieeexplore.ieee.org/document/10600829/ Instituto Brasileiro de Geografia e Estatística (IBGE). Produção agrícola municipal – culturas temporárias e permanentes. SIDRA: Tabela 5457. Disponível em: https://sidra.ibge.gov.br/tabela/5457. Acesso em: 5 mar. 2023. Bruno Henrique dos Santos Silva, Werônica Meira de Souza. Estudo Climatológico Da Cidade De Taquaritinga Do Norte – PE Para A Produção De Café Arábica Coffea arabica. BJAS [Internet]. 24 de dezembro de 2022 [citado 8 de março de 2025];4(2). Disponível em: https://www.journals.ufrpe.br/index.php/BJAS/article/view/5390 Simões M. ASPECTOS CLIMÁTICOS DO ESTADO DE MINAS GERAIS. Revista Brasileira de Climatologia. 2015;17. Graph Graph 1 is available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Graph1.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 19 Jan, 2026 Reviews received at journal 15 Jan, 2026 Reviewers agreed at journal 08 Jan, 2026 Reviews received at journal 08 Dec, 2025 Reviewers agreed at journal 19 Nov, 2025 Reviewers invited by journal 19 Nov, 2025 Editor assigned by journal 09 Nov, 2025 Submission checks completed at journal 09 Nov, 2025 First submitted to journal 08 Nov, 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-8064468","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":547557213,"identity":"9980df0c-e351-4245-b78f-bc6c4f0780cb","order_by":0,"name":"Rogério William Fernandes Barroso¹","email":"data:image/png;base64,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","orcid":"","institution":"Instituto Federal de Educação, Ciência e Tecnologia do Sul de Minas Gerais","correspondingAuthor":true,"prefix":"","firstName":"Rogério","middleName":"William Fernandes","lastName":"Barroso¹","suffix":""},{"id":547557214,"identity":"bd269336-c9dd-4ee9-ab8a-35032063f0ab","order_by":1,"name":"Adriano Bortolotti Silva²","email":"","orcid":"","institution":"Universidade José do Rosário Vellano","correspondingAuthor":false,"prefix":"","firstName":"Adriano","middleName":"Bortolotti","lastName":"Silva²","suffix":""},{"id":547557215,"identity":"c70a56ac-ef01-47b6-b1a1-abee4d7e2c4c","order_by":2,"name":"Lucas Eduardo Oliveira³","email":"","orcid":"","institution":"Instituto Federal de Educação, Ciência e Tecnologia do Sul de Minas Gerais","correspondingAuthor":false,"prefix":"","firstName":"Lucas","middleName":"Eduardo","lastName":"Oliveira³","suffix":""}],"badges":[],"createdAt":"2025-11-08 14:08:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8064468/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8064468/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97136202,"identity":"a4d9a3b2-2e15-4264-91f0-941e864c5e4b","added_by":"auto","created_at":"2025-12-01 09:56:01","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1292335,"visible":true,"origin":"","legend":"","description":"","filename":"TheoreticalandAppliedClimatology.docx","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/3d879196e055698ddb227a86.docx"},{"id":97136378,"identity":"8adc01f6-f4c4-461f-b714-8116cd269c89","added_by":"auto","created_at":"2025-12-01 09:56:29","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5544,"visible":true,"origin":"","legend":"","description":"","filename":"b7023e7903e54efdb265cb4434da190d.json","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/f69872f63a63bbf30b82ce37.json"},{"id":97135796,"identity":"3e82511f-dd28-40ba-a707-ad8466325ab9","added_by":"auto","created_at":"2025-12-01 09:53:43","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":86367,"visible":true,"origin":"","legend":"","description":"","filename":"b7023e7903e54efdb265cb4434da190d1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/10f8658d347b97860b80a6eb.xml"},{"id":96938732,"identity":"ae825e7e-2619-46de-849f-740400ef3603","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30846,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/ba5b2d26836003bea34972c9.png"},{"id":96938741,"identity":"72189421-a476-42fe-8a47-2f230ef0f66f","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":746971,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/82b55daebb19ee67115b4c47.jpeg"},{"id":96938736,"identity":"535c7de3-e455-4b33-8894-2349d13e6b20","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":187670,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/387237e70fe3e1f27d103f8f.png"},{"id":96938731,"identity":"b1968238-857c-4fe4-829a-9092a43db430","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":71322,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/ba47c9ffce4ad2c22d198abf.png"},{"id":96938734,"identity":"dd637b9c-0007-441d-a725-75d09ec131de","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57418,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/f5ab4b704b749f36e7189ed4.png"},{"id":97135713,"identity":"856eb76a-0663-412d-8463-7ac88cb63258","added_by":"auto","created_at":"2025-12-01 09:53:02","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14214,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/e9f4df770346fde2256e36bb.png"},{"id":96938743,"identity":"4c4858d3-373f-432d-af19-b199c3dacc32","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":159167,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/d4e6fb2ebf8ad16ec3c22347.png"},{"id":97135782,"identity":"5230ef1c-3a61-4791-a1c5-ff0f4bdba935","added_by":"auto","created_at":"2025-12-01 09:53:37","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55498,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/cfed6e2d1befb7981266d3d5.png"},{"id":97135670,"identity":"72aff664-81b3-48dc-a8f0-5951b38e17b1","added_by":"auto","created_at":"2025-12-01 09:52:45","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18660,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/6992ad9a0ac7fe25ba5f79ce.png"},{"id":96938738,"identity":"125a0051-c98b-45cb-8111-2c2939c208db","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19358,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/7696862a4ac3caa4cc1f5472.png"},{"id":97136313,"identity":"5ea51f9d-cfb9-4fc5-8655-fa59086c4d09","added_by":"auto","created_at":"2025-12-01 09:56:22","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":84634,"visible":true,"origin":"","legend":"","description":"","filename":"b7023e7903e54efdb265cb4434da190d1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/14305b248014e69c1493180c.xml"},{"id":96938740,"identity":"36061a14-655a-440b-9be9-171b853be442","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93711,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/0b73b1fdca5dd51528f6e4ed.html"},{"id":96938725,"identity":"1d278059-7c3f-4486-9ca7-c4d884431696","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":89149,"visible":true,"origin":"","legend":"\u003cp\u003eExample of an MLP architecture with N neurons in the input layer and in the hidden layer, two others in the hidden layer, and finally, two neurons in the output layer.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/f19690beba4250dfdc847a33.png"},{"id":97136732,"identity":"0d6f3450-dafc-44c9-97f2-bcafca686e48","added_by":"auto","created_at":"2025-12-01 09:56:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":288624,"visible":true,"origin":"","legend":"\u003cp\u003eResults of the models subjected to the parameters that best fit the data.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/f56e82ff0341250bcdf3777a.png"},{"id":97136831,"identity":"a77c6112-0d13-4e86-84b7-5ca682fa1029","added_by":"auto","created_at":"2025-12-01 09:57:02","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":196602,"visible":true,"origin":"","legend":"\u003cp\u003eMeteorological data from NASA POWER by Mesoregion of Minas Gerais between 2011 and 2021.\u003c/p\u003e\n\u003cp\u003ePrepared by: Rogério W. F. Barroso and Lucas E. A. de Oliveira. Data source: NASA POWER (2023).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/4f07183b69ea63facb7b7096.png"},{"id":96938727,"identity":"846e80c8-c2ab-46b2-a282-ed71753468e2","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":69875,"visible":true,"origin":"","legend":"\u003cp\u003eSoil characteristics and properties by Mesoregion of Minas Gerais between 2011 and 2021, extracted from the SoilGrids system. Prepared by: Rogério W. F. Barroso and Lucas E. A. de Oliveira. Data source: Poggio et al. (2021).\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/8487b15adc826df3008b2053.png"},{"id":97248381,"identity":"04771018-1ac9-47ce-8c22-41f7462e2ea1","added_by":"auto","created_at":"2025-12-02 12:56:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1199666,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/9a30a1ee-9803-4a45-ad11-c477fd76a287.pdf"},{"id":96938729,"identity":"b0fc4994-6430-49b5-a59c-4ae2752cc94a","added_by":"auto","created_at":"2025-11-27 17:25:33","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":71805,"visible":true,"origin":"","legend":"","description":"","filename":"Graph1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8064468/v1/d8221ad82ee5dc4317a68bc3.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Artificial Intelligence for Predicting Arabica Coffee Yield Based on Climate and Soil in the State of Minas Gerais, Brazil","fulltext":[{"header":"1. INTRODUCTION","content":"\u003cp\u003eCoffee is an agricultural commodity that generates wealth and holds an important position in both national and global economies. Brazil stands out as the world\u0026rsquo;s largest producer and exporter of Arabica coffee, accounting for 42% of global production.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e According to ABIC (2023), coffee is present in Brazilian households, showing a 1.64% increase in consumption in 2023 compared to the same period of the previous year.\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e This increase in consumption reflects the importance of coffee farming in the economic and social development of the country, significantly contributing to income generation.\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe states of Minas Gerais, S\u0026atilde;o Paulo, and Esp\u0026iacute;rito Santo stand out for their extensive production areas, totaling 1,422,715 hectares in 2024, according to EMBRAPA (2024). In particular, the state of Minas Gerais stands out as the largest coffee-producing state in Brazil, with 1,108,211 hectares.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e In 2023, the gross production value of Arabica coffee in Brazil was R\u003cspan\u003e$\u003c/span\u003e 36,804,111.19, with the Southeast region accounting for R\u003cspan\u003e$\u003c/span\u003e 34,274,655.55 and the state of Minas Gerais reaching R\u003cspan\u003e$\u003c/span\u003e 26,539,536.46, representing more than 70% of the gross production value of the entire country, according to a report by MAPA (2023).\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThese data are consistent with the favorable conditions for coffee cultivation in Brazil. ACEVES-NAVARRO et al. (2020) reported that the optimal temperatures for coffee production range from 19 to 22\u0026deg;C, which can be found in the state of Minas Gerais, characterized by an ideal combination of altitude, temperature, and precipitation, providing the necessary conditions for Arabica coffee cultivation (DAMATTA et al., 2018; DOORENBOS \u0026amp; KASSAM, 1979; TAVARES et al., 2018; LOREN\u0026Ccedil;ONE et al., 2021). Climatic variability directly influences the plant, resulting in fluctuations in yield, as evidenced by L\u0026auml;derach et al. (2017) and Monteiro et al. (2017).\u003csup\u003e\u003cspan additionalcitationids=\"CR6 CR7 CR8 CR9 CR10\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe research also considered soil composition, a factor that directly impacts coffee production. According to Kouadio et al. (2018), soil pH, organic matter and nutrient content, texture, and structure can significantly influence fertility, and the absence of any of these aspects may result in less fertile soils, reduced water retention and aeration, decreased infiltration, increased erosion, and hindered supply of essential nutrients to plants.\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eA wide range of factors can affect yield, and different tools can support the understanding of results. These tools are artificial intelligence models, which represent an accessible alternative, mathematically less complex, facilitating process automation in computerized systems.\u003c/p\u003e\u003cp\u003eTheir application has been widely used in the agronomic field for yield forecasting of crops such as coffee and soybean, proving to be an ally in planning and decision-making at harvest time by farmers, enabling a deeper analysis of large datasets and variables that may affect coffee yield more simply.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eConsidering that studies have not used temporal data on climate, soil, and yield in machine learning models, this research aimed to investigate the application of Temporal Difference (TD), Multilayer Perceptron (MLP), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), Gradient Boosting Regressor (GBR), and K-Nearest Neighbors (KNN) models to understand the relationships among soil properties, climate variables, and Arabica coffee yield data obtained from 200 municipalities in the state of Minas Gerais.\u003c/p\u003e"},{"header":"2. MATERIAL AND METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003e2.1. Study area\u003c/h2\u003e\n \u003cp\u003eThe research was conducted in the state of Minas Gerais, using historical data series over 10 years (2011\u0026ndash;2021) from 200 municipalities with high Arabica coffee production, according to IBGE (2023).\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e These municipalities were divided into twelve mesoregions for individualized analysis of climate and soil variables.\u003c/p\u003e\n \u003cp\u003eClimate data were obtained from the National Aeronautics and Space Administration/Prediction of Worldwide Energy Resources (NASA/POWER) platform, a tool validated in several regions of the world through studies comparing surface data, demonstrating its accuracy with respect to maximum, minimum, and mean temperature parameters in Egyptian territories.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003cp\u003eThe Application Programming Interface (API) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://power.larc.nasa.gov/api/pages/\u003c/span\u003e\u003c/span\u003e) was used as the tool to retrieve the following variables: air temperature at 2 meters (T2M), maximum (T2M_MAX) and minimum (T2M_MIN) temperature at the same height, relative humidity at 2 meters (RH2M), precipitation (PRECTOTCORR), surface pressure (PS), and wind speed at 10 meters (WS10M).\u003c/p\u003e\n \u003cp\u003eSoil data were retrieved from the SoilGrids platform, a global information system containing spatial predictions for various soil properties (TIFAFI et al., 2017), through the SoilGrids REST API 2.0 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://rest.isric.org/soilgrids/v2.0/docs\u003c/span\u003e\u003c/span\u003e). The following input variables were considered for this study: clay content (Clay), total nitrogen (Nitrogen), organic carbon density (OCD), organic carbon stocks (OCS), soil pH (PHH2O), and soil organic carbon content in the fine earth fraction (SOC). The data were organized by mesoregions of the state of Minas Gerais.\u003c/p\u003e\n \u003cp\u003eThe state of Minas Gerais has a wide variety of soils and microclimates, with coffee cultivation being practiced across all of them, particularly in the South/Southwest, Zona da Mata, and Tri\u0026acirc;ngulo Mineiro mesoregions, which are responsible for most of the coffee production. The IBGE Automatic Recovery System \u0026ndash; SIDRA allowed us to perform the geographical analysis of the producing mesoregions.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003e2.2. Artificial intelligence models\u003c/h2\u003e\n \u003cp\u003eThe use of artificial intelligence for yield prediction has been widely applied, especially in the field of agronomic sciences, for example, coffee yield forecasting, soybean yield forecasting through machine learning, soybean yield prediction, and sugarcane yield estimation.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003cp\u003eHowever, no literature is found when it comes to comparing several predictive models for estimating Arabica coffee yield in various municipalities of Minas Gerais. According to James et al. (2013), there are several models aimed at classifying based on probabilities, one of which is the k-nearest neighbors (KNN) method. Given a positive integer K and a test observation x₀, the KNN classifier identifies the K points closest to x₀ in the training set, represented by N₀, and then estimates the conditional probability for class j as the fraction of points in N₀ whose response values are equal to j:\u003csup\u003e22\u003c/sup\u003e\u003c/p\u003e\n \u003cdiv id=\"Equa\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eAccording to G. Biau \u0026amp; E. Scornet (2016), the Random Forest (RF) method can be described as M regression trees randomly distributed, creating several decision trees. This method works on the simple and effective logic of \u0026ldquo;divide and conquer,\u0026rdquo; splitting the data into smaller parts, building a random tree model for each segment, and subsequently integrating these models to form a consolidated prediction. This supervised algorithm can be applied to various types of problems, whether with small or large datasets, requiring only a small number of parameters.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003cp\u003eAnother widely used classification model is the Support Vector Machine (SVM). According to James et al. (2013), SVM gained great popularity in the 1990s due to its strong performance and distinct approach of finding a hyperplane that separates the data in the best possible way\u0026mdash;something not present in classical approaches such as logistic regression and linear discriminant analysis.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003cp\u003eThe Multilayer Perceptron (MLP) is used in complex scenarios that require supervised learning algorithms to employ multiple hyperplanes and address problems that are not linearly separable. This model allows for multiple input and output layers, in addition to hidden layers that determine the number of neurons, which in turn define the model\u0026rsquo;s hyperplanes, enabling more flexible and nonlinear separations. In contrast, SVM applies a single optimal hyperplane to separate the classes.\u003c/p\u003e\n \u003cp\u003eTemporal Difference (TD) is a concept of reinforcement learning. This technique consists of updating state values based on the difference between the current prediction and the observed future reward. According to Sutton and Barto (1998), this method addresses prediction problems using experiences accumulated over time.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e The algorithm generates an immediate target at time step t\u0026thinsp;+\u0026thinsp;1 and performs an update using the observed reward R\u003csub\u003e\u003cem\u003et+1\u003c/em\u003e\u003c/sub\u003e and the estimate V(\u003cem\u003eS\u003c/em\u003e\u003csub\u003e\u003cem\u003et+1\u003c/em\u003e\u003c/sub\u003e) of the next state. The simplest TD method performs this update without considering the value of states beyond the next one, adjusting directly after each transition, known as \u003cem\u003eTD(0)\u003c/em\u003e:\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e\n \u003cp\u003eExtreme Gradient Boosting (XGBoost) is an efficient and widely used algorithm based on the boosting method. According to Alshboul et al. (2022), it was developed as an enhancement of the gradient boosting method, being mainly applied to regression and classification tree problems.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e The concept of boosting forms its foundation, combining the prediction of weak learners with additive training methods to build a more robust and accurate model. Mathematically, the objective function of XGBoost is expressed as:\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e\n \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:l\\left({y}_{i},{y}_{i}^\\right)\\)\u003c/span\u003e\u003c/span\u003e represents the loss function between the prediction and the true value, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varOmega\\:\\left({f}_{k}\\right)\\)\u003c/span\u003e\u003c/span\u003e is a regularization term that penalizes model complexity to avoid overfitting by capturing noise and irrelevant patterns, which would impair performance on new data.\u003c/p\u003e\n \u003cp\u003eFinally, the Gradient Boosting Regressor (GBR), similar to XGBoost, is an ensemble learning model based on the boosting method. According to Otchere et al. (2021), GBR consists of an iterative collection of decision trees organized sequentially, where each tree learns from the errors of the previous one to improve model accuracy. Mathematically, a GBR with M trees can be represented as:\u003c/p\u003e\n \u003cdiv id=\"Equd\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:hm\\left(xj\\right)\\)\u003c/span\u003e\u003c/span\u003e represents each weak learner tree, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{m}\\)\u003c/span\u003e\u003c/span\u003e is the weight assigned to each tree, and M is the total number of trees. This process allows GBR to progressively reduce residual error, making it a powerful method for regression and classification problems.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e2.3. Model evaluation metrics\u003c/h2\u003e\n \u003cp\u003eRegression analysis was performed to understand the relationships among soil properties, climate variables, and Arabica coffee yield data obtained from 200 municipalities in the state of Minas Gerais and to verify which model achieved the highest accuracy. The difference between the actual data points and the best-fit line produced by the algorithm is considered the model error. The model error with multiple data points was determined using the following criteria, characterized by JAMES et al. (2013): coefficient of determination (R\u003csup\u003e2\u003c/sup\u003e), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Squared Error (MSE). These metrics are essential to ensure proper model evaluation and were applied in this study to select the best machine learning algorithm.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003c/div\u003e"},{"header":"3. RESULTS","content":"\u003cp\u003eThe predictive models were subjected to several parameters to identify which achieved the best results among them (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The performance analysis demonstrates the accuracy of each machine learning model for predicting Arabica coffee yield in Minas Gerais and highlights the specific adequacy of each technique to the characteristics of the dataset used.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u0026ndash; Parameters defined and tested in each machine learning algorithm used to predict Arabica coffee yield in Minas Gerais.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eParameter\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTested Values\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003eTD\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003emax_depth\u003c/p\u003e\u003cp\u003emin_samples_split\u003c/p\u003e\u003cp\u003emin_samples_leaf\u003c/p\u003e\u003cp\u003erandom_state\u003c/p\u003e\u003cp\u003ecriterion\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4, 5, 6, 7, 8, 9, 10\u003c/p\u003e\u003cp\u003e2,5,10,15,20\u003c/p\u003e\u003cp\u003e1,5,15,20\u003c/p\u003e\u003cp\u003e10\u003c/p\u003e\u003cp\u003efriedman_mse, poisson,\u003c/p\u003e\u003cp\u003esquared_error, absolute_error\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eMLP\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003ehidden layer sizes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(9; 10; 9); (7; 8; 10); (10; 10; 10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003emax iter\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e100; 1000; 1500; 7000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003ecross-Validation\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003elearning rate init\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.01; 5; 10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003elearning rate\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003econstant; adaptive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eActivation\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003erelu; tanh; logistic\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eSolver\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eadam, sgd\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eAlpha\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.01; 5; 10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003erandom state\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eSVM\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eKernel\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003erbf; linear; poly; sigmoid\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.1; 50; 100; 1000; 10000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003e\u0026Eacute;psilon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.1; 1; 4; 5; 10; 100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eGamma\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eauto; scale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eDegree\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1; 2; 3; 4; 7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003ecoef0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0; 1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eRF\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003emax depth\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003enone, 5, 10, 15\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003en estimators\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10; 200; 500; 1000; 10000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003erandom state\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eXGBOOST\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003emax depth\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1;3;5; 10; 100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eAlpha\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.1, 20; 200; 2000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003en estimators\u003c/p\u003e\u003cp\u003elearning_rate\u003c/p\u003e\u003cp\u003eobjective\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e100; 200;300; 3000\u003c/p\u003e\u003cp\u003e0.1, 0.01, 0.05\u003c/p\u003e\u003cp\u003ereg:squarederror\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eGBR\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003elearning_rate\u003c/p\u003e\u003cp\u003en estimators\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.01, 0.05, 0.1\u003c/p\u003e\u003cp\u003e10; 100; 500; 1000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003emax depth values\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003enone; 5; 10; 100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003erandom state values\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eKNN\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003en neighbors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e\u003cp\u003e3;5;7;9;11\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWeights\u003c/p\u003e\u003cp\u003eMetric\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e\u003cp\u003euniform, distance\u003c/p\u003e\u003cp\u003eeuclidean, Manhattan\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThe models tested include RF, TD, SVM, KNN, GBR, XGBoost, and MLP, each with an optimal hyperparameter configuration adjusted to maximize prediction accuracy. For this purpose, the scikit-learn library method was employed.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u0026ndash; Application of the optimal parameters in the models and their results.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"11\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e\u003cp\u003eTraining\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c11\" namest=\"c7\"\u003e\u003cp\u003eTesting\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e\u003cp\u003eModel performance measures\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMAPE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eR\u0026sup2;\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eRMSE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMSE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eMAE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eMAPE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eR\u0026sup2;\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c9\"\u003e\u003cp\u003eRMSE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c10\"\u003e\u003cp\u003eMSE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c11\"\u003e\u003cp\u003eMAE\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.057\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.87\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e34.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.04\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.177\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e65.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e4287.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e5.50\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.105\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4.26\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e18.13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.209\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.996\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e55.68\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e3100.26\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e5.44\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eSVM\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e0.990\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.999\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.05\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.00\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.04\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1.126\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.999\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.05\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e\u003cb\u003e0.00\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e\u003cb\u003e0.04\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.0004\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e2094.086\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.764\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e467.87\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e218905.09\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e276.64\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGBR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.487\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.62\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.70\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e3.214\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.996\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e55.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e3025.46\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e3.84\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eXGBOOST\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e25.877\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e7.34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e53.83\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e3.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e152.569\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.996\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e60.91\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e3710.21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e8.61\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMLP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e82.385\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.52\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e210.86\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e10.63\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e214.021\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e21.14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e446.99\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e12.55\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThe RF model showed excellent accuracy in both training and testing, with R\u003csup\u003e2\u003c/sup\u003e close to 1 in both cases, indicating a high capacity of the model to explain the variance in the data. This model also presented low error values (MAE and RMSE) in both training and testing, standing out as one of the most robust models for this type of data.\u003c/p\u003e\u003cp\u003eThe TD model also showed high accuracy, with performance similar to RF during training, but with a slight increase in error values during testing, which may indicate the onset of overfitting. Nevertheless, it remains a solid choice due to its interpretability and ease of implementation.\u003c/p\u003e\u003cp\u003eThe SVM, known for its effectiveness in high-dimensional spaces, achieved virtually zero errors in training, reflecting an almost perfect adaptation to the training data, and maintained this characteristic in the test set. This performance suggests that the linear SVM model is extremely efficient for this specific dataset, possibly due to the linearity between the explanatory variables and the response.\u003c/p\u003e\u003cp\u003eIn contrast, the KNN model, despite showing perfect performance in training, exhibited a large discrepancy in testing, with extremely high errors, indicating a significant generalization issue for new data, perhaps due to its sensitivity to outliers or the distribution of the test data.\u003c/p\u003e\u003cp\u003eBoth GBR and XGBoost, as boosting models, achieved good results, with low error metrics during training and acceptable values in testing, demonstrating good generalization ability and robustness. These models are known for their efficiency in complex and heterogeneous datasets, which is typical of agricultural data.\u003c/p\u003e\u003cp\u003eLastly, the MLP, a type of neural network, showed the greatest disparity between training and testing results, indicating considerable overfitting despite its high training accuracy.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eSVM achieved the most accurate prediction of Arabica coffee yield among all the models analyzed. This result corroborates the findings of Kouadio et al. (2018), Kumar et al. (2018), and Kilic et al. (2024), who applied the model in agricultural contexts, providing farmers with robust and practical tools for forecasting yields, diseases, and pest infestations, integrating them into farm management systems, optimizing resource use, and maximizing production.\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe same dataset used to predict coffee yield in the municipality of Muzambinho in 2021 was employed to perform tests with the trained model, where the average production yield was 1,200 kilograms per hectare, according to IBGE \u0026ndash; SIDRA (2023). The prediction was 1,386 kilograms per hectare when submitted to the trained algorithm, a difference of 186 kilograms per hectare for the municipality.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e"},{"header":"4. DISCUSSION","content":"\u003cp\u003eIt is necessary to evaluate a range of factors related to production, given the influence and impact each one has on the results. Some studies address the yield associated with climate\u003csup\u003e\u003csup\u003e[1]\u003c/sup\u003e\u003c/sup\u003e and the biophysical aspects of the soil\u003csup\u003e\u003csup\u003e[2]\u003c/sup\u003e\u003c/sup\u003e separately. Therefore, we considered both aspects related to production in this study. Climate characteristics, such as temperature, affect coffee production. According to Loren\u0026ccedil;one et al. (2021), the ideal temperatures for coffee cultivation range from 19 to 22 \u0026deg;C, while leaf yellowing is observed above 25 \u0026deg;C, favoring the emergence of pests. Figures 2 and 3 show that all mesoregions of the state of Minas Gerais present suitable temperatures. However, rainfall conditions are not homogeneous. According to SILVIA \u0026amp; SOUZA (2022), yield was greater in regions with higher rainfall rates, as observed in Campo das Vertentes, Metropolitan Region of Belo Horizonte, West of Minas Gerais, South/Southeast of Minas Gerais, Tri\u0026acirc;ngulo Mineiro/ Alto Parana\u0026iacute;ba, and Zona da Mata.\u003csup\u003e9,30\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eSoil plays an important role in the coffee production process, and its fertility is a highly dynamic process influenced by several factors, including nutrient availability, such as pH, organic matter, soil texture, and nitrogen, among others. According to Kouadio et al. (2018), soils with low pH are less favorable for nutrient absorption and are consequently less fertile. This pattern is standardized across all regions (Figure 4). Organic matter, which is important for increasing water retention, soil infiltration, aeration, and reducing erosion, in addition to supplying nutrients to plants, shows slight discrepancies among the regions. Ultisols, characterized by higher clay content across all mesoregions, predominate in the state of Minas Gerais.\u003csup\u003e12\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eClimate influences coffee yield, which can be confirmed in Graph 1, presenting the average production yield by mesoregion. According to Sim\u0026otilde;es et al. (2015), the south of Minas Gerais is the rainiest, with an average annual precipitation of around 1,600 mm. This aligns with Graph 1, which exhibits that the South/Southwest of Minas Gerais is the region with the largest number of municipalities included in the analysis, showing a good distribution of production, even with some municipalities above and below the general average, followed by the Tri\u0026acirc;ngulo Mineiro/Alto Parna\u0026iacute;ba, which also includes highly productive municipalities.\u003csup\u003e31\u003c/sup\u003e\u003c/sup\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eThe graph shows large differences in yield among the regions of Minas Gerais. Some regions, such as the South/Southwest of Minas Gerais and Tri\u0026acirc;ngulo Mineiro, stood out for their higher yield and consistency, while others, such as Vale do Rio Doce and Vale do Mucuri, exhibited characteristics that were not favorable for coffee cultivation. Variations in climate can be observed, although soil properties do not vary substantially, as illustrated in the histograms.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e\u003csup\u003e[1]\u003c/sup\u003e\u003c/sup\u003e LOREN\u0026Ccedil;ONE, Jo\u0026atilde;o Antonio \u003cem\u003eet al\u003c/em\u003e, PREVIS\u0026Atilde;O DA PRODUTIVIDADE DO CAF\u0026Eacute; COM BASE EM DADOS AGROCLIM\u0026Aacute;TICOS E APRENDIZAGEM DE M\u0026Aacute;QUINA / FORECASTING COFFEE YIELD BASED ON AGROCLIMATIC DATA AND MACHINE LEARNING, \u003cstrong\u003eInternational Journal of Environmental Resilience Research and Science\u003c/strong\u003e, v. 3, n. 1, 2021.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e\u003csup\u003e[2]\u003c/sup\u003e\u003c/sup\u003e KOUADIO, Louis \u003cem\u003eet al\u003c/em\u003e, Artificial intelligence approach for the prediction of Robusta coffee yield using soil fertility properties, \u003cstrong\u003eComputers and Electronics in Agriculture\u003c/strong\u003e, v. 155, p. 324\u0026ndash;338, 2018.\u003c/p\u003e"},{"header":"5. CONCLUSION","content":"\u003cp\u003eThis study conducted a comparative analysis of several machine learning models, including SVM, Random Forest, Decision Tree, KNN, GBR, XGBoost, and MLP, to predict Arabica coffee yield in Minas Gerais, using six soil datasets (clay, total nitrogen, carbon density, carbon stock, pH, and carbon content) and seven climate datasets (2-meter air temperature, maximum and minimum 2-meter temperature, 2-meter humidity, rainfall, surface pressure, and 10-meter wind speed) as predictor variables, and coffee yield as the target variable. SVM proved to be the most efficient tested model when incorporating soil and climate data. The potential of integrating artificial intelligence algorithms with different crops is reinforced with this study, supporting farmers\u0026rsquo; decision-making and improving harvest planning, as well as contributing to public policies aimed at coffee producers.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eR.W.F.B. conceived and conducted the research, performed data collection, and wrote the main manuscript text.A.B.S. supervised the study and contributed to the interpretation of the results and critical review of the manuscript.L.E.O. provided methodological guidance, assisted in data validation, and reviewed the final version of the manuscript.All authors reviewed and approved the final manuscript.The Corresponding Author, R.W.F.B., was responsible for all communications with the journal and ensuring compliance with Springer\u0026rsquo;s authorship policies.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eI thank the Coordination for the Improvement of Higher Education Personnel (CAPES) for the financial support, the Federal Institute of Education, Science, and Technology of Southern Minas Gerais \u0026ndash; Muzambinho Campus for the support and encouragement, and Jos\u0026eacute; do Ros\u0026aacute;rio Vellano University (UNIFENAS) for the opportunity to enroll in the Doctoral Program. Finally, and no less importantly, I thank my advisors, Prof. Dr. Lucas Eduardo de Oliveira and Prof. Dr. Adriano Bortolotti da Silva, for their constant support, dedication, and guidance throughout this journey.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eEmbrapa (2024), Sum\u0026aacute;rio executivo: Produ\u0026ccedil;\u0026atilde;o mundial de caf\u0026eacute; estimada para safra 2023\u0026ndash;2024 totaliza 171,4 milh\u0026otilde;es de sacas de 60 kg, Embrapa, publicado em 18 abril 2024, dispon\u0026iacute;vel em: https://www.embrapa.br/busca-de-noticias/-/noticia/88547345/producao-mundial-de-cafe-estimada-para-safra-2023-2024-totaliza-1714-milhoes-de-sacas-de-60kg (acesso em: 23 ago. 2024).\u003c/li\u003e\n \u003cli\u003eIndicadores da Ind\u0026uacute;stria de Caf\u0026eacute; (2023), ABIC. Dispon\u0026iacute;vel em: https://estatisticas.abic.com.br/estatisticas/indicadores-da-industria/indicadores-da-industria-de-cafe-2023/. Acesso em: 17 mai. 2024. [Internet]. ABIC. [citado 16 de maio de 2024]. Dispon\u0026iacute;vel em: https://estatisticas.abic.com.br/estatisticas/indicadores-da-industria/indicadores-da-industria-de-cafe-2023/\u003c/li\u003e\n \u003cli\u003eFassio, L. H. and Silva, A. E. S. (2015), \u0026lsquo;Import\u0026acirc;ncia econ\u0026ocirc;mica e social do caf\u0026eacute; Conilon\u0026rsquo;, in: Livro Caf\u0026eacute; Conilon 2007, Instituto Capixaba de Pesquisa, Assist\u0026ecirc;ncia T\u0026eacute;cnica e Extens\u0026atilde;o Rural (Incaper), Esp\u0026iacute;rito Santo, pp. 34\u0026ndash;49. Em.\u003c/li\u003e\n \u003cli\u003eVBP_10_23.pdf [Internet]. [citado 16 de maio de 2024]. Dispon\u0026iacute;vel em: http://www.consorciopesquisacafe.com.br/images/stories/noticias/2021/2023/outubro/VBP_10_23.pdf\u003c/li\u003e\n \u003cli\u003eAceves-Navarro LA, et al. Impact of climatic change on the adaptation of coffee (Coffea arabica L.) crops in Tabasco, Mexico. Agroproductividad. 2020;13(4):53\u0026ndash;58.\u003c/li\u003e\n \u003cli\u003eDamatta FM, et al. Physiological and agronomic performance of the coffee crop in the context of climate change and global warming: a review. J Agric Food Chem. 2018;66(21):5264\u0026ndash;5274.\u003c/li\u003e\n \u003cli\u003eDOORENBOS, J.; KASSAM, A. H. Yield response to water. [s.l: s.n.].\u003c/li\u003e\n \u003cli\u003eTavares PDS, et al. Climate change impact on the potential yield of Arabica coffee in southeast Brazil. Reg Environ Change. 2018;18(3):873\u0026ndash;83.\u003c/li\u003e\n \u003cli\u003eLoren\u0026ccedil;one JA, Aparecido LEDO, Loren\u0026ccedil;one PA, Moraes JRDSCD. PREVIS\u0026Atilde;O DA PRODUTIVIDADE DO CAF\u0026Eacute; COM BASE EM DADOS AGROCLIM\u0026Aacute;TICOS E APRENDIZAGEM DE M\u0026Aacute;QUINA / FORECASTING COFFEE YIELD BASED ON AGROCLIMATIC DATA AND MACHINE LEARNING. Intern Journ Env Res Res Sci [Internet]. 5 de maio de 2021 [citado 16 de maio de 2024];3(1). Dispon\u0026iacute;vel em: https://e-revista.unioeste.br/index.php/ijerrs/article/view/26255\u003c/li\u003e\n \u003cli\u003eL\u0026auml;derach P, Ramirez-Villegas J, Navarro-Racines C, et al. Climate change adaptation of coffee production in space and time. Clim Change. 2017;141:47\u0026ndash;62. doi:10.1007/s10584-016-1788-9.\u003c/li\u003e\n \u003cli\u003eMonteiro JEB de A, et al. Modeling of corn yield in Brazil as a function of meteorological conditions and technological level. Pesq Agropec Bras. 2017;52(3):137\u0026ndash;48.\u003c/li\u003e\n \u003cli\u003eKouadio L, Deo RC, Byrareddy V, Adamowski JF, Mushtaq S, Phuong Nguyen V. Artificial intelligence approach for the prediction of Robusta coffee yield using soil fertility properties. Computers and Electronics in Agriculture. dezembro de 2018;155:324\u0026ndash;38.\u003c/li\u003e\n \u003cli\u003eTorsoni GB, De Oliveira Aparecido LE, Dos Santos GM, Chiquitto AG, Da Silva Cabral Moraes JR, De Souza Rolim G. Soybean yield prediction by machine learning and climate. Theor Appl Climatol. fevereiro de 2023;151(3\u0026ndash;4):1709\u0026ndash;25.\u003c/li\u003e\n \u003cli\u003eDe Oliveira Aparecido LE, De Souza Rolim G, Camargo Lamparelli RA, De Souza PS, Dos Santos ER. Agrometeorological Models for Forecasting Coffee Yield. Agronomy Journal. janeiro de 2017;109(1):249\u0026ndash;58.\u003c/li\u003e\n \u003cli\u003eGuimar\u0026atilde;es EDS. Aprendizado de M\u0026aacute;quina aplicado \u0026agrave; predi\u0026ccedil;\u0026atilde;o da produtividade da cultura da soja utilizando dados de clima e solo [Internet] [Mestrado em Matem\u0026aacute;tica, Estat\u0026iacute;stica e Computa\u0026ccedil;\u0026atilde;o]. [S\u0026atilde;o Carlos]: Universidade de S\u0026atilde;o Paulo; 2020 [citado 16 de maio de 2024]. Dispon\u0026iacute;vel em: https://www.teses.usp.br/teses/disponiveis/55/55137/tde-09062020-123106/\u003c/li\u003e\n \u003cli\u003eIBGE (2023), Sistema IBGE de Recupera\u0026ccedil;\u0026atilde;o Autom\u0026aacute;tica - SIDRA: produ\u0026ccedil;\u0026atilde;o agr\u0026iacute;cola municipal: tabelas. Minas Gerais. Dispon\u0026iacute;vel em: https://sidra.ibge.gov.br/tabela/5457. Acesso em: 5 mar. 2023.\u003c/li\u003e\n \u003cli\u003eStackhouse PW, Westberg D, Hoell JM, Chandler WS, Zhang T. Prediction of worldwide energy resource (POWER) \u0026ndash; agroclimatology methodology (1.0 latitude by 1.0 longitude spatial resolution). NASA Tech Rep. 2015.\u003c/li\u003e\n \u003cli\u003eAboelkhair, H., Morsy, M. and Afandi, G. E. (2019), \u0026lsquo;Assessment of agroclimatology NASA POWER reanalysis datasets for temperature types and relative humidity at 2 m against ground observations over Egypt\u0026rsquo;, Advances in Space Research, Vol. 64, No. 1, pp. 129\u0026ndash;42.\u003c/li\u003e\n \u003cli\u003eTifafi M, Guenet B, Hatt\u0026eacute; C. Large Differences in Global and Regional Total Soil Carbon Stock Estimates Based on SoilGrids, HWSD, and NCSCD: Intercomparison and Evaluation Based on Field Data From USA, England, Wales, and France. Global Biogeochemical Cycles. 2018;32(1):42\u0026ndash;56.\u003c/li\u003e\n \u003cli\u003eSim\u0026otilde;es, J. C. and Pelegrini, D. F. (2010), Diagn\u0026oacute;stico da Cafeicultura Mineira \u0026ndash; Regi\u0026otilde;es Tradicionais: Sul/Sudoeste de Minas, Zona da Mata, Tri\u0026acirc;ngulo Mineiro/Alto Parana\u0026iacute;ba. EPAMIG, Belo Horizonte. (EPAMIG. S\u0026eacute;rie Documentos, n. 46).\u003c/li\u003e\n \u003cli\u003eBarbosa LAF, Pedronette DCG. ESTUDO COMPARATIVO ENTRE DIFERENTES REGRESSORES PARA ESTIMAR PRODUTIVIDADE DE CANA-DE-A\u0026Ccedil;\u0026Uacute;CAR.\u003c/li\u003e\n \u003cli\u003eJames, G., Witten, D., Hastie, T. and Tibshirani, R. (2013), An Introduction to Statistical Learning: With Applications in R. Springer, New York.\u003c/li\u003e\n \u003cli\u003eBiau G, Scornet E. A random forest guided tour. TEST. 1\u003csup\u003eo\u003c/sup\u003e de junho de 2016;25(2):197\u0026ndash;227.\u003c/li\u003e\n \u003cli\u003eSutton, R. S. and Barto, A. G. (1998), Reinforcement Learning. MIT Press, Cambridge, MA.\u003c/li\u003e\n \u003cli\u003eAlshboul, O., Shehadeh, A., Almasabha, G. and Almuflih, A. S. (2022), \u0026lsquo;Extreme Gradient Boosting-Based Machine Learning Approach for Green Building Cost Prediction\u0026rsquo;, Sustainability, Vol. 14, No. 11, 6651. https://doi.org/10.3390/su14116651.\u003c/li\u003e\n \u003cli\u003eApplication of gradient boosting regression model for the evaluation of feature selection techniques in improving reservoir characterisation predictions.\u003c/li\u003e\n \u003cli\u003eKumar S, Mishra S, Khanna P, Pragya. Precision Sugarcane Monitoring Using SVM Classifier. 2018 [citado 9 de mar\u0026ccedil;o de 2025]; Dispon\u0026iacute;vel em: https://arxiv.org/abs/1803.09413\u003c/li\u003e\n \u003cli\u003eKilic I, Yaman O. Classification Method of Plants with Support Vector Machine (SVM) Using Local Binary Gaussian Model (LBGP) for Smart Agriculture. Em: 2024 32nd Signal Processing and Communications Applications Conference (SIU) [Internet]. Mersin, Turkiye: IEEE; 2024 [citado 9 de mar\u0026ccedil;o de 2025]. p. 1\u0026ndash;4. Dispon\u0026iacute;vel em: https://ieeexplore.ieee.org/document/10600829/\u003c/li\u003e\n \u003cli\u003eInstituto Brasileiro de Geografia e Estat\u0026iacute;stica (IBGE). Produ\u0026ccedil;\u0026atilde;o agr\u0026iacute;cola municipal \u0026ndash; culturas tempor\u0026aacute;rias e permanentes. SIDRA: Tabela 5457. Dispon\u0026iacute;vel em: https://sidra.ibge.gov.br/tabela/5457. Acesso em: 5 mar. 2023.\u003c/li\u003e\n \u003cli\u003eBruno Henrique dos Santos Silva, Wer\u0026ocirc;nica Meira de Souza. Estudo Climatol\u0026oacute;gico Da Cidade De Taquaritinga Do Norte \u0026ndash; PE Para A Produ\u0026ccedil;\u0026atilde;o De Caf\u0026eacute; Ar\u0026aacute;bica Coffea arabica. BJAS [Internet]. 24 de dezembro de 2022 [citado 8 de mar\u0026ccedil;o de 2025];4(2). Dispon\u0026iacute;vel em: https://www.journals.ufrpe.br/index.php/BJAS/article/view/5390\u003c/li\u003e\n \u003cli\u003eSim\u0026otilde;es M. ASPECTOS CLIM\u0026Aacute;TICOS DO ESTADO DE MINAS GERAIS. Revista Brasileira de Climatologia. 2015;17.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Graph","content":"\u003cp\u003eGraph 1 is available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Artificial intelligence, Yield, Predictive models, Arabica coffee","lastPublishedDoi":"10.21203/rs.3.rs-8064468/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8064468/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eCoffee is an agricultural commodity of great economic importance for Brazil and worldwide. In this context, the country also stands out as the largest producer and exporter of Arabica coffee. Furthermore, coffee consumption in Brazil grew by 1.64%, reflecting its social and economic relevance. The state of Minas Gerais leads national production, with more than 1 million hectares under cultivation and accounting for over 70% of the gross production value of Arabica coffee in 2023. Factors such as climate and soil directly contribute to this result, impacting yield. This research aimed to study temperature variations and optimal soil conditions so that farmers can plan and forecast yield expectations year after year. In this context, this study applied predictive artificial intelligence models, such as TD, MLP, RF, XGBoost, SVM, GBR, and KNN, including in the analyses variables such as climate, soil, and coffee yield, using data from 200 municipalities in Minas Gerais. The goal was to support farmers in estimating their production, contributing to better harvest planning and optimizing production costs.\u003c/p\u003e","manuscriptTitle":"Artificial Intelligence for Predicting Arabica Coffee Yield Based on Climate and Soil in the State of Minas Gerais, Brazil","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-27 17:25:28","doi":"10.21203/rs.3.rs-8064468/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-01-19T18:49:59+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-15T13:31:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"148261525825510490729705685855399068199","date":"2026-01-08T19:47:37+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-08T15:12:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"142377768564533613117092458908343171325","date":"2025-11-19T13:12:14+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-19T12:51:26+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-10T04:49:19+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-10T04:47:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"Theoretical and Applied Climatology","date":"2025-11-08T14:05:14+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"9e177a8e-f2ec-47ca-b790-6808cdf9de05","owner":[],"postedDate":"November 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-02-23T13:24:27+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-27 17:25:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8064468","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8064468","identity":"rs-8064468","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}