Decoding Urban Emotions: Exploring the Association between Urban Built Environment and Residents' Emotional Health Using Interpretable Machine Learning Models

Decoding Urban Emotions: Exploring the Association between Urban Built Environment and Residents' Emotional Health Using Interpretable Machine Learning Models | 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 Decoding Urban Emotions: Exploring the Association between Urban Built Environment and Residents' Emotional Health Using Interpretable Machine Learning Models meng Cai, xiaoyin Zhang, xue Gong This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7774299/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract With the advancement of urbanization and the improvement of living standards, residents' demands for living environments and quality of life are increasingly high. The built environment not only shapes the daily living space of residents but also significantly impacts their emotional well-being. This study, from a grid-based micro perspective, integrates geospatial big data and social media data from Weibo (Chinese Twitter), employing an interpretable machine learning model that combines XGBoost and SHAP to conduct an in-depth analysis of the complex relationship between the built environment and residents' emotional health in Nanning City. The study finds that, first, residents' emotional health exhibits distinct temporal and spatial distribution characteristics; second, transportation stations and green spaces are the two environmental variables that most significantly affect residents' emotional health; third, there is a nonlinear relationship and threshold effect between built environmental elements and residents' emotional health, indicating that the impact of built environment elements on emotional health tends to stabilize or reverse after reaching certain thresholds; fourth, there are interactive effects among different built environmental elements, suggesting that certain combinations of environmental elements may have a more pronounced impact on residents' emotional health. These findings also highlight the importance of considering multidimensional environmental characteristics and their interactions in urban planning to enhance residents' emotional health and achieve sustainable urban development. The innovation of this study lies in the combination of Weibo big data and geospatial big data, using an interpretable machine learning model to precisely capture the distribution characteristics of residents' emotions at a small-scale spatial level and explore the complex relationship with the 5D built environment. This provides a basis for optimizing the layout of urban built environment elements from the perspective of residents' emotions and holds significant theoretical and practical significance for urban planning and health management. Urban built environment Weibo social media data Emotional health Interpretable machine learning X-minute city life circle Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction 1.1 Research background With the continuous advancement of global urbanization, it is predicted that by 2050, nearly 70% of the global population will reside in urban areas (Heilig, 2012). Cities, as the primary spaces for population aggregation, serve not only as engines of economic growth but also as hubs for daily life, leisure, and entertainment. However, the "big city disease" and unhealthy lifestyles resulting from rapid urbanization have imposed a substantial burden on residents' physical and mental health (Gong et al., 2012). Among these issues, mental health problems are more likely to be overlooked due to their uncertainty and intangibility. Sustainable cities and communities (SDG11) have been identified by the United Nations as one of the key agendas for 2030, emphasizing the improvement of the built and social environments in urban planning and community development to enhance residents' health and well-being. In China, the concept of a "healthy city" has also emerged as one of the significant goals to be achieved by 2030. This goal necessitates the integration of health considerations into every aspect of urban planning, construction, and management, thereby promoting residents' health and bolstering urban sustainability. Emotion is a product of the interaction between individuals and their environment. Psychologist Broek (2013) identified six basic emotional categories: anger, disgust, fear, happiness, sadness, and surprise. From these fundamental emotions, other non-basic emotions can be derived, such as relaxation, oppression, anxiety, and security. While emotional research is prevalent in psychology, sociology, and medicine, the growing interdisciplinary exchanges have led geography, architecture, and related fields to increasingly focus on the study of emotion and space. Emotional health denotes a stable and positive emotional state that confers health benefits (Guo & Du, 2023). Existing research indicates that emotions are closely linked to individuals' physical health, mental health, and social adaptation. Particularly, positive emotions can reduce the risk and likelihood of illness, enhance physical and mental well-being, and improve quality of life by influencing cognitive processes and encouraging healthier behaviors (Steptoe et al., 2009; Steptoe & Diez, 2008). Moreover, positive emotions are associated with broader social connections and support. For instance, positive emotions can expand individuals' interpersonal resources, such as friendships and social networks, which in turn provide social support and mutual aid, helping individuals effectively navigate various life challenges and difficulties (Guo & Wang, 2007). The urban built environment mainly refers to the physical spaces such as urban functional spaces and transportation systems constructed through human planning, design, construction and renovation, including housing, schools, commercial areas, roads, etc. (Handye et al., 2005). Initial research on the built environment identified evaluation indicators across three dimensions—density, diversity, and design—establishing a 3D framework for assessing the built environment (Cervero & Kockelman, 1997). Subsequently, two additional dimensions—destination accessibility and distance to transportation hubs—were incorporated into this framework, resulting in a more comprehensive 5D indicator system for the built environment. This system, which includes density, diversity, design, destination accessibility, and distance to transportation stations (Ewing et al., 2010), has become the principal basis for contemporary evaluations of the built environment. The built environment not only defines the daily living spaces of residents but also significantly influences their emotional states. Specifically, the walkability of the built environment directly impacts residents' daily activities and emotions. A walkable community environment can markedly enhance physical activity levels, thereby boosting positive emotions and reducing stress (Hematian & Ranjbar, 2022). Moreover, accessibility to service facilities and entertainment venues is crucial, as these spaces provide essential social and leisure areas that help alleviate stress and increase happiness, thus positively affecting residents' emotions (Säumel et al., 2016). Additionally, a well-designed community fosters neighborly communication and cooperation, enhances community cohesion, and instills a greater sense of safety and belonging among residents. Conversely, issues such as poor walking experiences, walking fears, or spatial disorganization can heighten anxiety, diminish happiness, and even deter healthy behaviors (Aram et al., 2019; Li et al., 2015). 1.2 Literature review Research on the urban built environment and emotional health can be traced back to the concept of emotional geography proposed by British scholars Anderson and Smith in 2001. They formally introduced the concept of emotional geography, exploring the interaction between human emotions—such as joy, anger, sorrow, and happiness—and various spatial environments. This marked a shift in the study of human perception and the urban environment from purely subjective psychological domains to a broader social spatial dimension (Anderson & Smith, 2001). Currently, with the improvement of urban residents' quality of life and the transformation of urban construction from a stage of incremental expansion to one focusing on quality enhancement, the relationship between human behavior and the built environment has garnered increased attention. The connection between emotional activities and urban space has also become a key research focus. Numerous previous studies have explored the interaction between urban space and public sentiment, yielding meaningful results (He et al., 2024). For example, Chen et al. (2018) selected three streets with distinct characteristics in Beijing as research subjects to investigate the correlation between the landscape features of different streets and people's emotional responses. Han et al. (2024) compared the impact of urban-scale, community-scale, and residential-scale built environments on the mental health of the elderly, using Hefei City in Anhui Province, China, as a case study, and found that the urban-scale built environment had the highest relative importance. Chen et al. (2016) utilized emotion mapping analysis technology, combined with biofeedback instruments and GPS, to collect physiological indicators and geographical data from specific populations during real-life experiences. They studied the emotional responses elicited by the real-life environment and further explored the use of fractal measurement technology to construct an accurate emotion measurement model. Previous studies have shown that urban morphology, landscape environment and entertainment facilities have a significant impact on public sentiment (Chen et al. 2018; He et al., 2024). For example, researchers observed that adequate and high-quality urban greening and blue spaces such as lakes and rivers help trigger positive emotions (Huai & Van, 2022), in contrast, adverse environmental conditions such as noise, congestion and heat island effects, can lead to negative emotions (Holly et al., 2015). However, the researchers found that the relationship between urban environment and public sentiment was not comprehensively consistent. For example, many studies have shown that road networks bring many conveniences to people's lives and travel, but Perez et al. (2022) observed that excessive road density will bring negative effects such as noise disturbances, environmental pollution and traffic congestion. Similarly, many studies have reported that high building density tends to cause negative emotions (He et al., 2022), however, Huang et al. observes that high building density in Hong Kong can also promote positive emotions among residents (Huang et al., 2023), because There may be a synergy between building density and other factors such as infrastructure quality and amenities. It can be seen from this that the impact of urban spatial environment on residents' lives and emotions is not always a linear and single direct relationship, but may have opposite effects or mixed effects with other factors after exceeding a certain threshold. Therefore, it is necessary to deeply analyze whether there is a complex relationship between the built environment and emotional health. In addition, existing studies often describe residents' emotional characteristics at macro-levels such as cities, streets, and communities, exploring the relationship between individual emotions and spatial environments (Liu et al., 2020; Chen et al., 2018). In contrast, research on the spatial and temporal distribution characteristics of average emotions within grid units and their relationship with the built environment, particularly at the micro-scale of 15-minute living circles, is still in its exploratory phase. The 15-minute city living circle is a people-oriented urban planning concept based on time urbanism, envisioning that residents can easily access basic services and amenities, such as convenience stores, healthcare facilities, schools, and parks, within a 15-minute walk or cycle from their homes (Murgante et al., 2024; Moreno et al., 2021). As the global urbanization process deepens and people's demands for living environments and quality of life increase, the 15-minute city living circle concept is gaining widespread attention. An increasing number of cities worldwide are integrating this concept into urban planning and construction. For example, the C40 City Initiative, involving 100 major cities across multiple countries, emphasizes the importance of implementing 15-minute cities. Cities such as Milan, Paris, Bogota, Portland, and Melbourne have taken actions to reconfigure the built environment to promote intelligent, low-carbon, and sustainable community development (C40 Cities Climate Leadership Group, 2020; Allam et al., 2022). Since China introduced the 15-minute city living circle planning guide in Shanghai, many Chinese cities, including Jinan, Guangzhou, Hangzhou, and Suzhou, have incorporated this concept into their planning documents and gradually promoted it in national urban renewal efforts (Nie, 2024). Therefore, it is essential to deepen the discussion of the relationship between the built environment and residents' health at the micro-geographical level of the 15-minute city living circle. Furthermore, current research often focuses on describing residents' emotional characteristics at the macro level, such as cities, streets, and communities, exploring the relationship between individual emotions and the spatial environment (Liu et al., 2020; Chen et al., 2018). In contrast, research examining the spatiotemporal distribution characteristics of residents' emotions within micro-grid units, especially from the perspective of living circles, and their nonlinear relationships, threshold effects, and interactive effects with the built environment, is still in the exploratory phase. The concept of living circles is a people-oriented urban planning philosophy, based on the concept of temporal urbanism, envisioning that residents can easily access basic services and conveniences, such as convenience stores, healthcare facilities, schools, and parks, within approximately 5, 10, or 15 minutes' walk or bike ride from their residential areas (Murgante et al., 2024; Moreno et al., 2021). With the deepening of the global urbanization process and the increasing demands for living environments and quality of life, the concept of living circles is gradually receiving widespread attention. An increasing number of cities worldwide are incorporating the living circle philosophy into urban planning and construction, such as the C40 Cities Initiative, which involves 100 major cities from several countries, emphasizing the importance of implementing the 15-minute city, and cities like Milan, Paris, Bogotá, Portland, and Melbourne, have already implemented actions focused on reconfiguring the built environment to promote intelligent, low-carbon, and sustainable community development (C40 Cities Climate Leadership Group, 2020; Allam et al. 2022). Since Shanghai introduced the 15-minute CLC planning guidelines, many Chinese cities, including Jinan, Guangzhou, Hangzhou, and Suzhou, have incorporated this concept into their planning documents and are gradually promoting its practice in urban renewal nationwide (Nie, 2024). Therefore, it is necessary to deepen the exploration of the relationship between the built environment and residents' health at the micro geographic level of living circles. Finally, existing research primarily relies on questionnaires and interviews to gather data. However, due to limitations in scale and spatial and temporal precision, these methods struggle to fully capture the relationship between large areas, large samples, and the real-time comprehensive urban built environment and residents' emotions. With the rise of social media applications, people have increasingly become accustomed to sharing their daily lives and expressing their opinions on these platforms. Against this backdrop, location-based social network (LBSN) data, such as that from Weibo and Twitter, has emerged and rapidly provided a wealth of real-time public sentiment data. These emerging social media big data sources offer potential solutions to the shortcomings of traditional public opinion research methods (Shan et al., 2022). Despite ongoing controversies regarding privacy, representation, and other issues (Murthye et al., 2015), social media data is progressively supplanting traditional survey data as “social sensors” to enhance our understanding of real-world social phenomena. 1.3 Research content and contributions To summarize, This paper first collects user posting information on Weibo social media in Nanning City, Guangxi, from January to December 2023, and employs the SnowNLP library in Python to analyze the emotion values of the cleaned posts. It also establishes a 15-minute city living circle scope to create independent grids, calculates the average emotional values of residents in each grid, and analyzes their spatial and temporal distribution characteristics. Secondly, this paper constructs a 5D environmental index system, including key indicators such as population density, building density, functional mix, road intersection density, road network density, greening rate, commercial service centers, and public transportation station density. Based on this, the gradient decision tree (XGBoost) machine learning method is used to identify the contribution ranking and characteristic effects of different built environmental factors on residents' emotional health. Finally, the paper discusses the complex nonlinear relationships and threshold effects between various built environmental factors and residents' emotional health. The contributions of This paper are as follows: First, by constructing a grid unit based on the 15-minute living circle, this study can more precisely capture and analyze the distribution characteristics of residents' emotions in small-scale spaces and examine the subtle effects of the built environment within the 15-minute living circle on residents' emotional health. Second, this paper utilizes social media big data and built environment big data to expand the data sources, enhance the timeliness and comprehensiveness of the research, and provide a more accurate and real-time perspective for evaluating and understanding the impact of the urban built environment on residents' emotional health. This, in turn, offers robust data support and a scientific basis for the development of urban planning and public health interventions. Third, based on the XGBoost machine learning model and the SHAP interpretability framework, we explore the nonlinear relationships and threshold effects between various built environmental elements and residents' emotions, thereby compensating for the limitations of traditional regression methods that rely solely on linear analysis. 2. Materials and Methods 2.1 Research area The research area of This paper primarily focuses on the central urban area of Nanning City, the capital of Guangxi Autonomous Region, located in western China. There are two main reasons for selecting Nanning City. Firstly, as the capital and economic and cultural hub of Guangxi Zhuang Autonomous Region, Nanning City exhibits a high degree of urbanization and population concentration. Its urban built environment and residents' living conditions are highly representative. Moreover, the significant population aggregation results in abundant and accessible social media data (such as Weibo), which can provide a wealth of residents' emotional health data for research purposes. Secondly, most prior studies have concentrated on developed cities, with fewer investigations into underdeveloped cities in less developed regions. As a city in the underdeveloped western part of China, Nanning can offer valuable insights and data support for other similar regions through this research, assisting urban planners and policymakers in better incorporating residents' emotional health into urban construction, optimizing the layout of the built environment, and enhancing residents' quality of life. Nanning City comprises 7 urban districts, 4 counties, and 1 county-level city. Among these, Qingxiu District, Xixiangtang District, Jiangnan District, Xingning District, Liangqing District, and Yongning District are the six main urban districts (Nanning Municipal Government, 2023), with the central urban area being the confluence of these districts. The research area was delineated based on the urban central street, as shown in Fig. 1 . According to the "Standards for Planning and Design of Urban Residential Areas," the walking distance for a 15-minute living circle should be 800–1100 meters to meet basic living needs (Ministry of Housing and Urban-Rural Development, 2018). Therefore, this study adopted an 800m×800m grid unit, independently dividing the research area into grids of this size. Subsequent statistics on built environmental factors, residents' emotional health, and corresponding data analysis were all conducted based on these grid units. 2.2 Research data 2.2.1 Urban built environment data The 5D built environment indicator system encompasses five dimensions: density (including population and building density), diversity (pertaining to the variety of urban functional land uses and activity types), design (the planning and design of blocks and buildings), distance (to transportation facilities and service centers), and accessibility (including the convenience of walking, cycling, and public transportation) (Ewing et al., 2010; Wang Enxu et al., 2024). Accordingly, This paper adopts the established 5D built environment indicator system to assess the built environment level of Nanning's central urban area. Drawing on existing research, the characteristics of Nanning's central urban area, and data availability, This paper selects nine indicators: population density, building density, mixed use of urban functional land, road network density, greening rate, commercial service centrality, indoor sports and leisure service centrality, and density of public and transportation stations in parks and green spaces (Table 1 ). Using ArcGIS 10.8, we calculated and extracted the built environment variables within each 800m×800m grid cell based on the selected indicators and built environment data. Figure 2 illustrates the spatial distribution of the nine types of built environment indicators, revealing significant spatial differentiation patterns across different levels of the built environment. Table 1 5D built environment indicators and sources Dimension index Calculation method Source density Population density Population number/grid area (people/m 2 ) China's 7th Census Data https://www.stats.gov.cn/sj/z xfb/202302/t20230203_1901080.html? Building density Building base area/grid area (m) Landsat8 satellite imagery https://earthexplorer.usgs.gov/ diversity Urban functional land mix Area ratio of various functional land in the grid OpenStreetMap https://openmaptiles.org/ design Road network density Total length of the road network in the grid/grid area(m2) OpenStreetMap https://openmaptiles.org/ Greening rate Normalized vegetation index NDVI mean in grid Geospatial Data Cloud https://www.gscloud.cn/ Destination accessibility Business Service Center The average core density of commercial service POI in the grid, such as clothing stores, supermarkets and restaurants, etc. Gaode Map Open Platform https://developer.amap.com/ ndoor sports and leisure service center Mean core density of POI in the grid sports and leisure service category, such as gyms and cinemas Gaode Map Open Platform https://developer.amap.com/ Park green space The mean core density of scenic spots and services in the grid, such as parks, green squares, etc Gaode Map Open Platform https://developer.amap.com/ Traffic stop distance Traffic site density The mean core density of traffic stations in the grid, including three categories: bus, subway and private car parking stations Gaode Map Open Platform https://developer.amap.com/ 2.2.2 Resident emotional health data The data utilized in This paper originates from Sina Weibo, one of China's largest social media platforms. In 2023, Weibo boasted 598 million users, akin to Twitter, which is globally widespread. It permits users to post text, images, videos, and links to document and share their lives, express emotions, and voice opinions on various events and issues. This study amassed 219,960 Weibo posts related to Nanning city, geotagged with location markers, published on the Sina Weibo platform from January to December 2023. Each post encompasses content, release time, location, authentication type, and author details. The dataset underwent desensitization, noise reduction, and removal of invalid samples from non-study areas, culminating in 188,103 Weibo posts. Subsequently, SnowNLP was employed to analyze and process the text content of these posts. SnowNLP is a Python-based method specifically designed for processing Chinese text, capable of accurately calculating the probability of a post's sentiment being positive, with an accuracy rate exceeding 80% (Zhang et al., 2018). Specifically, texts conveying joy, happiness, satisfaction, or admiration are categorized as "positive" emotions, whereas those expressing sadness, anger, disappointment, or depression are deemed "negative" emotions. The emotional probability value ranges from 0 to 1. Drawing on existing research and aligning with the content and data characteristics of this study, an emotional score of 0.6 and below is considered a low mood value, 0.6 to 0.8 is a medium mood value, and above 0.8 is a high value (Shan et al., 2022). Lastly, following He et al. (2024), emotional health value is defined as the average of emotional values from all Weibo users' posts within any 800m×800m grid cell to reflect the emotional health of residents in that area. Grids without any Weibo user posts are treated as missing values and excluded from subsequent data analysis.Table 2 shows the descriptive statistics of each variable. Table 2 Descriptive statistics of variables Variable Mean SD Min Max emotional health value 0.730 0.205 0 1 population density 0.006 0..007 0 0.034 building density 0.135 0.093 0 0.447 land use mix 0.012 0.010 0 0.030 road network density 0.008 0.005 0 0.031 NDVI 0.333 0.123 0.009 0.766 commercial service centrality 253.258 295.753 0 2623.23 indoor recreational centrality 11.137 13.679 0 102.031 green space centrality 1.775 3.425 0 24.533 transit station density 30.663 34.278 0 212.566 2.3 Model method This paper employs nine types of built environment elements as characteristic variables and residents' emotional health as the response variable to construct an interpretable model integrating XGBoost and SHAP to assess the impact on residents' emotional health. The XGBoost model, proposed by Chen et al. (2016), is an efficient algorithm that significantly improves upon the gradient boosting decision tree (GBDT). XGBoost is an ensemble machine learning model based on the gradient boosting algorithm, comprising multiple classification and regression trees (CARTs). Its fundamental principle involves continuously iterating to generate new decision trees that fit the residuals from the preceding tree model and summing the predicted values of these decision trees. Thus, the XGBoost model can be represented as the sum of the scores of k-CARTs, as illustrated below: $$\:\widehat{{\mathbf{y}}_{\mathbf{i}}}={\sum\:}_{\mathbf{k}=1}^{\mathbf{k}}{\mathbf{f}}_{\mathbf{k}}\left({\mathbf{x}}_{\mathbf{i}}\right)\:{(\mathbf{f}}_{\mathbf{k}}\in\:\mathbf{F})$$ 1 In the formula ( 1 ), \(\:\widehat{{\mathbf{y}}_{\mathbf{i}}}\) is the model predicted value of the i-th sample; \(\:\text{k}\) is number of trees; \(\:{\mathbf{x}}_{\mathbf{i}}\) is the eigenvector of the i-th sample; \(\:\mathbf{F}\:\) is a collection of all CART trees, \(\:{\mathbf{f}}_{\mathbf{k}}\) is the kth CART tree, that is, the predicted value of the x-th sample。 The overall idea of XGBoost is to combine the loss function and regularization term into a global loss function, and perform second-order Taylor expansion. Its objective function is: $$\:{\mathbf{S}}_{\mathbf{o}\mathbf{b}\mathbf{j}}\left(\varvec{\theta\:}\right)={\sum\:}_{\mathbf{i}=1}^{\mathbf{N}}\mathbf{L}({\mathbf{y}}_{\mathbf{i}},\widehat{{\mathbf{y}}_{\mathbf{i}}})+{\sum\:}_{\mathbf{j}=1}^{\mathbf{t}}\varvec{\Omega\:}\mathbf{ƒ}\left(\mathbf{j}\right)$$ 2 In the formula ( 2 ) \(\:{\sum\:}_{\text{i}=1}^{\text{N}}\text{L}({\text{y}}_{\text{i}},\widehat{{\text{y}}_{\text{i}}})\) represent the overall loss, express the model error, which is the difference between the true value of the sample and the predicted value; \(\:{\sum\:}_{\text{j}=1}^{\text{t}}{\Omega\:}\text{ƒ}\left(\text{j}\right)\) as a regularization term, it expresses the structural error of the model, i.e. the complexity of the regression tree, by using hyperparameters multiplied by the number and value of nodes to limit the complexity of the model. $$\:{\mathbf{S}}_{\mathbf{w}\mathbf{h}\mathbf{e}\mathbf{r}\mathbf{e}}\varvec{\Omega\:}\left(\mathbf{ƒ}\right)=\varvec{\gamma\:}\mathbf{T}+\frac{1}{2}\varvec{\lambda\:}{‖\varvec{\omega\:}‖}^{2}$$ 3 In the formula( 3 ) \(\:\varvec{\Omega\:}\left(\mathbf{ƒ}\right)\) is a regularization term of the model, used to reduce the overfitting problem and complexity of the model; \(\:\mathbf{T}\) is the number of leaf nodes; \(\:\varvec{\lambda\:}\) for the severity of punishment; \(\:\varvec{\omega\:}\) is output scores for leaf nodes; \(\:\frac{1}{2}\varvec{\lambda\:}{‖\varvec{\omega\:}‖}^{2}\) is the L2 modulus square of \(\:\varvec{\omega\:}\) .If the objective function of the model is smaller, the better the prediction effect of the model. The XGBoost model offers numerous advantages. Firstly, unlike traditional linear regression models, XGBoost does not presuppose a specific relationship between independent and dependent variables. Instead, it learns and automatically captures the complex nonlinear relationships by constructing decision trees, which traditional linear regression models cannot achieve. Secondly, compared to other tree-based models (such as GBDT), XGBoost is less sensitive to multicollinearity, missing values, outliers, and irrelevant variables. It can handle high-dimensional data effectively, is less prone to overfitting, and is highly adaptable to various datasets. It can process both discrete and continuous data without the need for normalization. 3. Result 3.1 The spatiotemporal distribution characteristics of residents' emotions in the research area 3.1.1 Time-change characteristics Based on the emotion values of all Weibo users posted within the research area, This paper first calculates the mean and median emotion values of Weibo users' posts on a monthly and weekly scale. Figure 3a illustrates the fluctuation trend of Weibo user sentiment in the study area throughout the year, with the median consistently above the mean and exhibiting similar fluctuation patterns. Specifically, the period from the end of the year to the beginning of the year, particularly January, marks a peak in user emotions, likely due to the festive atmosphere of holidays such as New Year's Day and the Chinese Spring Festival. During these holidays, people use social media platforms like Weibo to record and share joyful moments and New Year's greetings, thereby enhancing emotional value. In February, sentiment values dropped, with the mean and median falling to 0.731 and 0.843, respectively. This decline may be related to the adjustment of life routines following the Chinese Spring Festival holiday. The end of the holiday signifies the start of long-term work commitments and the separation from family, friends, and hometown, as well as the reluctance to leave the pleasant holiday times, all of which contribute to a decrease in residents' emotional values. From March to October, emotions remained relatively stable, with the mean fluctuating slightly between 0.740 and 0.752. After October, sentiment values rose again, with the mean and median in December reaching 0.799 and 0.914, respectively, marking another annual high. Figure 3b depicts the weekly emotional fluctuations of residents. The data indicates that emotional values are higher on rest days compared to weekdays, with Sunday recording the highest sentiment scores. This trend may be attributed to the "weekend effect." Following a week of work or study-related stress, individuals typically engage in more restful and relaxing activities over the weekend, such as family dinners, outdoor pursuits, or personal hobbies. These leisurely experiences render the weekend the most emotionally positive period of the week (Liu & Chai, 2001; Shan et al., 2022). Starting from Monday, emotion values gradually decline, reaching their nadir on Thursday. The particularly low emotional values observed on Thursday could be linked to the "midweek slump." People might feel somewhat fatigued by Thursday, having worked through most of the week without the imminent prospect of weekend relaxation and entertainment. This "neither here nor there" phase may temporarily dampen spirits (Yang et al., 2013). However, as the weekend approaches,emotion begins to recover on Friday and peaks once more on Saturday. Figure 3a month scale change chart of emotion value Fig. 3b weekly scale change chart of emotion value 3.1.2 Spatial distribution characteristics Based on the magnitude of emotional values and in conjunction with the research by Shan et al. (2022), this paper categorizes the grid units of the research area into three groups: ( 1 ) Areas with residents' emotional values of 0.8 and above are classified as high-emotion areas; ( 2 ) Areas with emotional values between 0.6 and 0.8 are designated as medium-emotion areas; ( 3 ) Areas with residents' emotional values below 0.6 are considered low-emotion areas. Figure 4 reveals that, generally, the distribution of emotional space in the study area is relatively fragmented, with a mixed distribution of high, medium, and low emotional values. It is noteworthy that the average emotional health value across the entire grid is 0.730, with a standard deviation of 0.205 (Table 2 ), indicating that the overall mood in Nanning is relatively positive and healthy. Figure 4 suggests a potential spatial correlation between urban environmental factors and residents' emotional health. For instance, compared to low and medium-emotion areas, high-emotion areas are more scattered and have a more pronounced distribution on the urban periphery. This may be because, as the city expands, new residential districts and commercial areas are developed outside the city. These areas may emerge as "new highlands" for emotional health due to factors such as low population density, less traffic congestion, and better natural environmental quality. This also reflects the "edge effect" of urban life: urban residents may feel alienated from crowded urban centers and may be more inclined to seek quieter and more private living spaces on the urban fringe to achieve a better emotional experience. This finding is consistent with a study conducted in San Francisco, USA. Based on sentiment analysis of the city's Twitter data, the study found that the periphery of cities with abundant vegetation resources exhibited significantly higher sentiment indexes than areas with limited vegetation resources (He et al., 2024), suggesting that individuals in areas with specific environmental conditions are more likely to develop positive emotions. 3.2 SHAP-based explanatory enalysis of the model Subsequently, this study employs the "xgboost" library in Python to analyze the intricate relationship between built environment variables and emotional indices. Initially, 80% of the samples are randomly designated as the training set to train the model, while the remaining 20% serve as the test set to assess model performance. Subsequently, the model's generalization ability is evaluated using a 5-fold cross-validation method to mitigate the risk of overfitting. Specifically, the training set is divided into five subsets, with four used for training in each iteration and the remaining subset used for validation. Following this, hyperparameter tuning is conducted using grid search, setting the learning rate to 0.1 (learning_rate: 0.1), selecting 50 decision trees (n_estimators: 50), with each tree having a maximum depth of 3 (max_depth: 3). Additionally, the subsample ratio for each tree is set to 0.7 (subsample: 0.7), and the ratio of randomly selected features is set to 0.8 (colsample_bytree: 0.8). The performance of the XGBoost model is assessed using mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE), with respective values of 0.036, 0.189, and 0.147. These results demonstrate the model's exceptional accuracy and predictive capabilities. After further exploring the complex relationship between the built environment and residents' emotional health using the XGBoost model, this paper employs the SHAP interpretation framework to visualize and explain the results of the XGBoost model. SHAP is a classic post-hoc attribution interpretation framework (Parsa et al., 2020) that not only outputs the importance ranking of variables but also more intuitively represents the local positive or negative effects of all samples of a variable on the model. Additionally, the output partial dependence plot (PDP) results can describe the degree of change of variables across different values, thereby enhancing the interpretability of the model. 3.2.1 Analysis of the importance of characteristic variables This paper integrates the SHAP interpretation framework to visualize and elucidate the analysis results of the XGBoost model. The SHAP mean value is utilized to illustrate the relative importance of characteristic variables, specifically the influence of various built-environment factors on residents' emotional health (Fig. 5a). It is evident that public transportation stations and green spaces are the most critical factors affecting emotional health, exerting a substantial impact on residents' emotions. Additionally, the density of the road network and the centrality of indoor entertainment facilities significantly contribute to emotional health. In contrast, building density, land use mixing, and the normalized vegetation index have relatively minor impacts, aligning mostly with anticipated and existing research findings. Although the Normalized Difference Vegetation Index (NDVI) demonstrated an important role, its impact was lower than expected and reported in relevant studies. This discrepancy may be attributed to Nanning, known as China's "Greentown," situated south of the Tropic of Cancer. With a humid subtropical monsoon climate, abundant sunshine, and plentiful rainfall, Nanning has an average annual temperature of around 21.6 degrees Celsius (China Weather Network, 2023). The vegetation is lush and evergreen year-round, and the urban greening coverage rate reaches 40.90% (Nanning Municipal Government, 2020). Streets, parks, and communities are lined with numerous trees and flowering plants, creating a uniform urban landscape. Consequently, residents accustomed to this pervasive green environment may experience minimal impact on their lives from variations in greening. 3.2.2 Local feature effects of feature variables On the basis of identifying the importance of feature variables, this paper uses SHAP feature summary plot to combine feature importance with feature effects, reflecting the strength, distribution, and direction of the influence of feature variables on predictor variables (Parsa et al., 2020), in order to reveal the complex relationship between various constructed environmental features and emotional health. As shown in Fig. 5b, on the vertical axis, each row represents a feature variable, namely the constructed built environment variable, and each point represents a sample. The redder the color of the point, the larger the value of the feature itself, and the bluer the color, the smaller the value of the feature itself. The horizontal axis represents the SHAP value of the feature variable, and the directions on both sides indicate positive or negative effects. If SHAP value > 0, it indicates that the feature variable, namely the built environment element, has a positive effect on residents' emotional health; SHAP value<0, This indicates that the feature variable has a negative effect on residents' emotional health |SHAP Value| represents the magnitude of the lift or decrease force on the predictor variable when this feature is used as a condition. For example, taking Land use mix as an example, the red line on the right side of Land use mix is longer than the blue line on the left side, that is, the red | SHAP Value | is greater than the blue, indicating that high land use mix can significantly affect residents' emotional health more than low land use mix. In addition, the Shapley values of high-value samples (red dots) in Land use mix are mostly positive, while the Shapley values of low value samples (blue dots) are mainly negative, indicating that high land use mix can promote residents' emotional health, while low land use mix can inhibit residents' emotional health. Although this graph provides valuable information, it cannot accurately quantify the impact of different values of the built environment on emotional health. Therefore, this study will use Partial Dependency Plot (PDP) to further elucidate the complex impact of built environment on emotional health within different value ranges. Figure 5a Analysis of the importance of characteristic variables Fig. 5b SHAP feature summary plot 3.2.3 Nonlinear relationship and threshold effect between built environment and residents' emotional health This paper further employs PDP to illustrate the impact of different built-environment characteristics on residents' emotional health under varying value conditions (Parsa et al., 2020). For each of the nine sets of graphs, the horizontal axis represents the feature variable, which is the constructed built-environment variable, and each point represents a sample. The larger the value, the higher the density value of the feature variable. The vertical axis represents the magnitude and direction of the impact of the built-environment variable on residents' emotional health (predicted value). If the impact is positive, it indicates that the feature has enhanced the predicted value; if the impact is negative, it indicates that the feature has diminished the predicted value. The absolute value represents the magnitude of the lifting or lowering force when this feature is used as a condition. From the partial dependency graph, it can be seen that each feature variable has an overall impact on residents' emotional health, as well as the direction and intensity of its specific values. Based on this, the mode of action of this feature variable can be analyzed (Lundberg et al., 2019).. From the results shown in the dependency graph in Fig. 6 , it can be observed that there is a significant nonlinear relationship and threshold effect between the built environment elements and emotional health. Specifically,to be more specific: Population Density: Within the range of 0-0.005, population density has a positive impact on emotional health; beyond 0.005, the influence shifts to negative and stabilizes after 0.01. This suggests that moderate population density may enhance emotional health due to increased economic activities and job opportunities, but excessively high density can have adverse effects due to overcrowding and resource scarcity (Gong et al., 2012; He et al., 2022). Additionally, as population density increases, competition for limited resources intensifies, potentially leading to heightened feelings of social injustice and psychological pressure (Želinský et al., 2021).Building Density: Within the range of 0-0.1, building density is positively correlated with emotional health; within 0.1–0.2, the impact is minimal; beyond 0.2, the influence becomes negative but weak. Thus, it can be determined that a building density of 0.2 is the critical point, where it contributes most positively to residents' emotional health. As building density continues to increase beyond 0.2, according to urban crowding theory, residents may begin to feel the congestion and stress of urban life, which could lead to a reduction in personal space, invasion of privacy, and competition for public resources (Shan et al., 2022), all of which could negatively impact residents' emotional health, although this negative effect is relatively weak. According to a study by Huang et al. (2024) in Hong Kong, it may be because areas of high density offer better infrastructure quality and convenient living conditions, which may compensate to some extent for the negative impacts of crowding (Huang et al., 2023).Land Use Mix: Above 0.02, land use mix is positively related to emotional health. This may be because increased mix implies greater convenience of services and facilities, thereby enhancing the quality of life (Murgante et al., 2024; Moreno et al., 2021).Road Network Density: Within the range of 0-0.003, road network density has a weakening effect on emotional health; between 0.003–0.01, emotional health levels improve; beyond 0.01, the impact turns negative. This may be because moderate road network density provides better travel convenience, but excessively high density could lead to negative effects due to traffic congestion and environmental issues.NDVI : Within the interval of 0-0.23, NDVI is positively correlated with emotional health; beyond 0.23, the impact diminishes. This could be because higher NDVI values indicate more green vegetation, offering psychological recovery spaces, but excessively high vegetation cover might bring about negative effects such as dim sunlight, insect bites, and pollen allergies (Maas et al., 2009). Commercial Service Centrality: Under most circumstances, the impact of commercial service centrality on residents' emotional health is weak. This may be attributed to the widespread adoption of e-commerce and online services in our country over the past decade, which has diminished residents' reliance on physical commercial services. However, when commercial service centrality is low, it has a positive effect on emotional health, likely associated with natural scenic areas. These areas typically exhibit lower commercial service centrality due to certain restrictions on commercial development, yet they may reap positive emotional health benefits because of access to beautiful natural scenery and a peaceful environment (Yuen et al., 2020).Indoor Recreational Centrality: In regions near zero, the absence of indoor recreational facilities is correlated with lower emotional health values; starting from 5, emotional health levels rise with the addition of indoor recreational facilities, stabilizing around 30. The lack of indoor entertainment facilities is associated with lower emotional health values, possibly because the absence of such facilities may lead to a loss of social capital and a dearth of cultural life (Aram et al., 2019), thereby limiting the richness of cultural experiences and spiritual life (Lin & Liu, 2024). Subsequently, as indoor recreational facilities increase, emotional health levels begin to plateau, likely because residents' needs for such facilities are gradually satisfied, and the marginal benefit of new facilities to emotional health diminishes. Moreover, residents may start to pay more attention to the personalization and quality of entertainment facilities, rather than just an increase in quantity. Green Space Centrality: In areas close to zero, the lack of green spaces is associated with lower emotional health values; starting from around 1, residents' emotional health increases with the addition of green spaces, stabilizing at around 10. The main findings indicate that in areas with low green space centrality, people lack recreational places and social opportunities. Subsequently, emotional health rise with the increase in green spaces, a phenomenon that can be attributed to the positive effects brought by nature and green areas. For instance, the renowned "20-minute park effect" proposed by Yuen et al. (2020) suggests that an increase in green spaces provides residents with more leisure and relaxation spots. Visiting urban parks, whether for a walk or just sitting, as long as there is intimate contact with nature for over 20 minutes, can significantly reduce stress hormone levels. However, after the critical value of 10, the improvement in emotional health begins to plateau, likely because residents' needs for green spaces are gradually met, and the marginal benefit of green spaces to emotional health starts to diminish.Transportation Station Density: Within the range of 0 to 10, there is a positive correlation between transportation station density and emotional health levels; beyond 10, a negative trend emerges. This could be because increased transportation stations enhance travel convenience, but an excessively high density, especially too many private car parking spots, may have negative impacts due to space occupation and environmental issues (Lunke, 2020). 3.2.4 The nonlinear relationship between the interaction effects of different built environment elements and residents' emotional health Finally, this study delves into the nonlinear impacts of the interactive effects between various built environment elements on residents' emotional health using PDP. Figure 7 utilizes SHAPley values to demonstrate the interactive effects between pairs of built environment variables. In each plot, the X-axis represents the variable of interest, with the coordinate value indicating the magnitude of that variable. The right Y-axis indicates the variable that interacts most strongly with the variable of interest, with its color representing the magnitude of that variable's value. Each point on the plot represents a sample point, with the color of the point corresponding to the value on the right Y-axis. Concurrently, the positive and negative SHAPley values on the left Y-axis signify the correlation between the feature variables and the target variable; values greater than zero indicate a synergistic effect, meaning that the two variables jointly promote an increase in the dependent variable; whereas values less than zero indicate an antagonistic effect, meaning that when the two feature variables act together, they have a negative impact on the target variable (such as emotional health). The distance on the Y-axis represents the significance of the correlation, with greater distance indicating stronger significance. Through these plots, we can more intuitively comprehend how different built environment characteristics interact with each other and subsequently influence residents' emotional health.For instance, the interaction effect plot between "Building Density" and "Population Density" reveals that as building density increases, the SHAP values show a downward trend, indicating that an increase in population in areas of high building density does not bring benefits to emotional health; instead, it may lead to antagonistic effects due to intensified competition for resources or overcrowding. The interaction effect plot between "Building Density" and "NDVI" shows that as the "Buildings" variable increases, the SHAP values also exhibit a downward trend, suggesting a potential antagonistic effect between the two, which means that an increase in NDVI may not significantly enhance residents' emotional health in areas with high building density. This finding suggests that merely increasing green space may not be sufficient to improve residents' emotional health, especially in areas where building density is already high. In contrast, the interaction effect plot between "Green Space Centrality" and "Commercial Service Centrality" shows that as "Commercial Service Centrality" increases, the SHAP values exhibit an upward trend, implying that improved accessibility to commercial facilities may positively affect residents' emotional health, particularly in environments with a higher degree of outdoor greenery. This synergistic effect may stem from commercial activities providing convenient living services for residents, while outdoor green spaces offer areas for relaxation and leisure, and the combination of the two may create a more livable environment. 4. Discussion This paper explores the complex relationship between the urban built environment and residents' emotional health from the perspective of a 15-minute living circle, using an interpretable machine learning model that integrates XGBoost and SHAP. Among the nine types of built environment features, transportation stations and green spaces were found to have the most significant impact on residents' emotional health. The article argues that transportation stations, such as subway stations, bus stations, and motor vehicle parking stations, as well as green spaces, such as parks, green squares, and natural landscapes, significantly impact residents' emotional health in the built environment because they touch on two fundamental dimensions of urban life: efficiency and resilience (vitality). Transportation stations are the lifeblood of daily urban life. They not only shorten physical distances but also alleviate time pressure, allowing residents to arrange their daily activities more flexibly (Li et al., 2020; Duan et al., 2023), thus demonstrating great importance. However, it is worth noting that overly dense traffic stations, especially the excessive increase in private stations, can lead to parking difficulties, squeezing and occupying space that originally belonged to greenery or transportation, which can have a significant negative effect. Green spaces are urban oases, and their positive effects on residents' emotional health are reflected in their restorative and therapeutic properties. They provide a way to reconnect with nature, help alleviate the stress and fatigue of urban life, stimulate positive emotions, promote relaxation and recovery, which is crucial for residents' emotional health (Woo et al., 2009; Xie et al., 2021). The urban built environment exhibits nonlinear relationships and threshold effects with residents' emotional health. This relationship indicates that there is a saturation point to the impact of increasing points of interest in the built environment on residents' emotional health; beyond this point, further increases have limited effects on enhancing emotional health and may even lead to negative outcomes. However, previous studies largely assumed a linear impact of the urban built environment on residents' emotional health, implying that changes in environmental factors would proportionally affect emotional health. The complexity of real-life scenarios suggests that such relationships are not always so straightforward and linear.For instance, when building density is between 0 and 0.1, there is a positive correlation between building density and emotional health values; between 0.1 and 0.2, there is almost no impact on emotional health; and beyond 0.2, building density has a negative effect on emotional health. The findings of this study can help urban planners and policymakers to more accurately understand the mechanisms by which the urban built environment affects residents' emotional health. By identifying and leveraging these nonlinear relationships and threshold effects, they can allocate resources rationally to maximize benefits in enhancing residents' emotional health. The interactive effects between environmental variables reveal which combinations of environmental characteristics may more effectively promote residents' emotional health, suggesting that urban planners need to consider multidimensional environmental characteristics and their interactions when designing urban spaces, to create a healthier and more livable urban environment.By taking these factors into comprehensive account, urban planning can not only enhance the material quality of life for residents but also promote their psychological well-being, thus achieving sustainable urban development. With the advancement of urbanization and the improvement of living standards, residents' demands for their living environment and quality of life are increasing. However, traditional urban planning, often dominated by functional zoning, has led to the centralized distribution of various living service facilities in cities. This results in longer travel times and higher costs for residents to access these facilities, thereby reducing their quality of life (Nie, 2024). The 15-minute living circle, where most urban residents conduct their daily activities, can significantly influence psychological feelings, activity willingness, and community atmosphere through the quality of its spatial elements. High-quality, humanized street design can effectively promote positive emotions (Fang, 2015; Murgante et al., 2014). In current urban areas, the increasing pace of life and work pressure, combined with uniform reinforced concrete buildings, not only leads to aesthetic and psychological fatigue but also has adverse effects on people's physical and mental health (Tsui, 2008). This phenomenon has prompted many urban residents to seek comfort and rejuvenation outside the city, with some even taking action to do so. For example, a short resignation letter that went viral in China a few years ago, "The world is so big, I want to see it," inspired many to follow suit. However, seeking restorative experiences or traveling outside the city for extended periods is not always feasible, as people need to work to make a living or stay behind to take care of their families (Lin & Liu, 2024). This raises an important question: How can individuals find opportunities for rest and rejuvenation in their daily lives? The Citywalk, which gained popularity in China in 2024, may provide a possible answer. Citywalk emphasizes exploring urban spaces on foot and immersing oneself in the local customs and development pulse of the city. This reflects a shift from escaping the city to immersing oneself in it. Urban residents no longer see the city solely as a place for work and life but begin to seek deeper connections with it, rediscovering and appreciating its charm through immersive urban experiences. This also indicates that cities are not only places for daily work but also the core space for residents' daily life and entertainment. Therefore, future urban construction and renewal should not only meet daily life and basic work needs but also consider promoting restorative and relaxation experiences, such as adding suitable natural spaces, leisure and entertainment facilities, and rich, diverse street designs, to help residents find a balance and source of vitality in the busy pace of urban life, thereby improving overall quality of life and happiness. 5. Conclusions This study used an interpretable machine learning model that integrates XGBoost and SHAP, combined with Weibo social media data and geographic big data, to conduct an in-depth analysis of the relationship between the built environment and residents' emotional health in Nanning city. The following conclusions can be drawn: Firstly, the overall emotional health level of urban residents in Nanning is relatively good. In terms of temporal distribution characteristics, residents' emotional health values exhibit significant regular changes within a week. Emotional values on rest days are higher than those on weekdays, with the highest values on Sundays, which may be related to increased leisure and relaxation activities during weekends. The lowest emotional point occurs on Thursdays, likely due to fatigue from continuous work. Over the course of a year, emotional health values peak during holidays such as New Year's Day and the Spring Festival, but decrease after holidays like in February and tend to fluctuate steadily in the subsequent months, indicating the positive impact of the holiday atmosphere on emotional health. Regarding spatial distribution characteristics, compared to low and median emotional areas, high emotional areas are more dispersed and have a more pronounced distribution in the peripheral areas of the city, while median and low emotional areas are concentrated and scattered in the central urban area. Secondly, regarding the relative importance and characteristic effects of built environment variables, research has found that traffic station density and green space centrality are the two most important factors affecting residents' emotional health. The convenience of transportation stations has significantly enhanced the convenience of daily commuting, reducing time costs and economic burdens. However, it is worth noting that the excessive increase of overly dense transportation stations, especially private stations, can have a significant negative effect. As urban "oases," green spaces provide residents with a way to reconnect with nature, helping to alleviate the pressure and fatigue of urban life and stimulate positive emotions. Thirdly, regarding the nonlinear relationship and threshold effects between the built environment and emotional health, the study finds that factors such as population density, building density, and land use mix are not simply linearly related to emotional health but exhibit threshold effects. At certain thresholds, emotional health levels are optimized, and beyond those thresholds, emotional health levels may decline. For instance, we observe a nonlinear relationship between transportation station density and emotional health. In areas with lower transportation station density (within the range of 0 to 10), there is a positive correlation between transportation station density and emotional health levels. When transportation stations exceed 10, there is a stable negative trend between transportation station density and emotional health levels. The interactive effects indicate which combinations of environmental characteristics may more effectively promote residents' emotional health, and these results also highlight the importance of considering multidimensional environmental features and their interactions in urban planning to achieve an improvement in residents' emotional health and sustainable urban development. Finally, there are several limitations in this study. Firstly, the Weibo data used in the study only reflects the emotions of users who use Weibo and like to post on Weibo, which makes the representative sample may be insufficient. Secondly, due to data limitations, the study was unable to control for other social attribute variables that may affect emotional health levels, such as economic status, education level, and age distribution. This resulted in the study only being able to explore the correlation between the built environment and emotional health to a certain extent, without delving into causal relationships. Thirdly, the study did not differentiate the distribution differences of built environment data in different regions of the city, such as the inner and outer urban areas. Due to differences in sensitivity and demand for built environment elements such as traffic density and green space among residents in different areas of the city, for example, suburban areas may be more sensitive to traffic station density, while residents in urban areas may be more concerned about green space. This difference needs to be more fully considered and analyzed in future research. Declarations Data availability statement ： The data that support the findings of this study are available from the corresponding author upon reasonable request. Disclosure of interest ： The authors declare no conflicts of interest in the research presented in this article. Funding information: This work was supported by the National Social Science Fund of China (No.22BSH109). Author Contribution meng Cai: Writing– review & editing, Validation, Supervision;Resources and funds;Xiaoyin Zhang: Writing–original draft, Visualization, Methodology. Investigation;xue Gong: Writing– review & editing, Validation, Supervision, Conceptualization. References Allam, Z., Bibri, S. E., Chabaud, D., et al. (2022). The ‘15-Minute City’ concept can shape a net-zero urban future. Humanities and Social Sciences Communications , 9 (1), 126. Anderson, K., & Smith, S. J. (2001). Emotional geographies. Transactions of the Institute of British geographers , 26 (1), 7–10. Aram, F., Solgi, E., & Holden, G. (2019). The role of green spaces in increasing social interactions in neighborhoods with periodic markets. Habitat International , 84 , 24–32. Bedimo-Rung, A. L., Mowen, A. J., & Cohen, D. A. (2005). The significance of parks to physical activity and public health: a conceptual model. American Journal of Preventive Medicine , 28 (2-supp-S2), 159–168. Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and design. Transportation research part D: Transport and environment, 2(3), 199–219. Chen Chongxian, Z., Danting, X., & Yu,&Li Shuhua. (2018). Research on the Impact of Urban Street Landscape Features on Human Emotional Health Urban architecture (9), 4. Chen, T., & Guestrin, C. (2016). XGBoost: a scalable tree boosting system. KDD'16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; 2016 Aug 13–17; San Francisco, California. Chen Zheng, W., Hangbin, & Yu Jue. (2016). & Empirical study on environmental reality perception for urban design Southern Architecture, (4), 10–14. China, E. (2023). Market Data Report https://www.100ec.cn/.2023 China Weather Network (2023). https://www.weather.com.cn/ C40 Cities Climate Leadership Group (2020). How to build back better with a 15-minute city. Available at: https://www.c40knowledgehub.org/s/article/How-to-build-backbetter-with-a-15-minute-city (accessed 30 November 2022). Duan Zhengyu, H., Xuan, Chen, & Chuan (2023). A Study on the Impact of Urban Residential Location and Travel Behavior on the Quality of Life of Commuters: A Case Study of Shanghai Modern Urban Studies (06), 88–93. Ewing, R., & Cervero, R. (2010). Travel and the built environment: A meta-analysis. Journal of the American planning association , 76 (3), 265–294. Fang Rong The spatial characteristics and planning design methods of living streets in 2015 Urban Issues 12:46–51. Gong, P., Liang, S., Carlton, E. J., Jiang, Q., Wu, J., Wang, L., & Remais, J. V. (2012). Urbanisation and health in China. The lancet , 379 (9818), 843–852. Guangxi Nanning Municipal People's Government Portal Website( (2023). https://www.nanning.gov.cn/ Guo Xiaoyan, W. Zhenhong (2007). Concept, Function, and Significance of Positive Emotions. Advances in Psychological Science , 15 (5), 810–815. Guo Yuting and Du Hongwu (2023). Analysis of Spatial Elements of Emotional Health oriented Lifestyle Streets: A Case Study of Guangzhou City Architecture and Culture (6), 122–124. Han Huiran, X., & Lingyi,&Yang Chengfeng. (2024). The impact of multi-scale built environment on the mental health of the elderly: an empirical study based on the extreme gradient boosting model in Hefei Geographic Research, 43 (6), 1502–1521. Handy, S., Cao, X., & Mokhtarian, P. (2005). Correlation or causality between the built environment and travel behavior? Evidence from Northern California. Transportation Research Part D: Transport and Environment , 10 (6), 427–444. He, D., Miao, J., Yi, L., Song, Y., Chen, L., & Liu, Y. (2022). Urban greenery mitigates the negative effect of urban density on older adults’ life satisfaction: evidence from Shanghai, China. Cities , 124 , 103607. He, P., Yu, B., Ma, J., et al. (2024). Exploring the non-linear relationship and synergistic effect between urban built environment and public sentiment integrating macro-and micro-level perspective: a case study in San Francisco. Frontiers in Psychology , 15 , 1276923. Heilig, G. K. Prospects:the 2022 Revision(United Nations,Department of Eeonomic and Social Affairs,Population Division,2012)[EB/OL] .[2023-08-15].http://esa.un.org/wpp/ppt/paa/PAA_2012_Heilig.pdf Hematian, H., & Ranjbar, E. (2022). Evaluating urban public spaces from mental health point of view: Comparing pedestrian and car-dominated streets. Journal of transport & health , 27 , 101532. Huai, S., & Van de Voorde, T. (2022). Which environmental features contribute to positive and negative perceptions of urban parks? A cross-cultural comparison using online reviews and Natural Language Processing methods. Landscape and Urban Planning , 218 , 104307. Huang, J., Cui, Y., Li, L., Guo, M., Ho, H. C., Lu, Y., & Webster, C. (2023). Re-examining Jane Jacobs’ doctrine using new urban data in Hong Kong. Environment and Planning B: Urban Analytics and City Science , 50 (1), 76–93. Hosford, K., Beairsto, J., & Winters, M. (2022). Is the 15-minute city within reach? Evaluating walking and cycling accessibility to grocery stores in Vancouver. Transportation Research Interdisciplinary Perspectives , 14, Article 100602. Jiang, Z., Wu, C., & Chung, H. (2025). The 15-minute community life circle for older people: Walkability measurement based on service accessibility and street-level built environment–A case study of Suzhou. China Cities , 157 , 105587. Jianxiao, L., Han, B. I., & Wang, M. (2020). Using multi-source data to assess livability in Hong Kong at the community-based level: A combined subjective-objective approach. Geography and sustainability , 1 (4), 284–294. Li Tao, W., & Jiaoe (2020). Happy River Research on the Differences in Inter city Travel Networks between Chinese Residents' Working Days and Holidays [J]. Journal of Geography , 75 (04), 833–848. Li Xinyang, H., & Zaisheng (2015). Yuan Yiqian, Zhao Lizhi&Xu Minjie Research on the suitability of community public spaces for urban elderly and children Urban Development Research (05), 104–111. Lin, Y. X., & Liu, Y. (2024).. Let the city heal you: environment and activity's distinct roles in leisure restoration and satisfaction. Cities , 154. Liu Zhilin, C. (2001). Yanwei The spatial structure of weekend leisure activities for Shenzhen citizens Economic Geography (04), 504–508. Lundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., … Lee,S. I. (2019). Explainable AI for trees: From local explanations to global understanding.arxiv preprint arxiv:1905.04610. Lunke, E. B. (2020). Commuters’ satisfaction with public transport. Journal of Transport & Health , 16 , 100842. Maas, J., Spreeuwenberg, P., Van Winsum-Westra, et al. (2009). Is green space in the living environment associated with people’s feelings of social safety? Environment and Planning A , 41 (7), 1763–1777. MacDonald, G. (2014). Bodies moving and being moved: Mapping affect in Christian Nold's Bio Mapping. Somatechnics , 4 (1), 108–132. Ministry of Housing and Urban-Rural Development of the People's Republic of China(2018). https://www.mohurd.gov.cn/ Moreno, C., Allam, Z., Chabaud, D., et al. (2021). Introducing the 15-Minute City: Sustainability, resilience and place identity in future post-pandemic cities. Smart Cities , 4 (1), 93–111. Murgante, B., Patimisco, L., & Annunziata, A. (2024).. Developing a 15-minute city: a comparative study of four italian cities-cagliari, perugia, pisa, and trieste. Cities, 146(Mar.), 1.1–1.22. Murthy, D., Gross, A., & Pensavalle, A. (2016). Urban social media demographics: An exploration of Twitter use in major American cities. Journal of Computer-Mediated Communication , 21 (1), 33–49. Nanning Municipal Government Website (2020). https://www.nanning.gov.cn/ Nie Jing (2024). Research on Urban Renewal Strategies Based on the 15 Minute Community Life Circle Concept Residential and Real Estate (16), 91–93. Parsa, A. B., Movahedi, A., Taghipour, H., Derrible, S., & Mohammadian, A. K. (2020). Toward safer highways, application of XGBoost and SHAP for real-time accident detection and feature analysis (Vol. 136, p. 105405). Accident Analysis & Prevention. Pérez-Crespo, L., Kusters, M. S., López-Vicente, M., Lubczyńska, M. J., Foraster, M., White, T., & Guxens, M. (2022). Exposure to traffic-related air pollution and noise during pregnancy and childhood, and functional brain connectivity in preadolescents. Environment International , 164 , 107275. Säumel, I., Weber, F., & Kowarik, I. (2016). Toward livable and healthy urban streets: Roadside vegetation provides ecosystem services where people live and move. Environmental Science & Policy , 62 , 24–33. Shan Zhuoran, A., Yuehui, Y., & Man (2022). etc Research on the spatiotemporal distribution characteristics and influencing factors of urban residents' emotions based on Weibo big data: taking the main urban area of Wuhan as an example [J]. Urban Development Research, 29 (8): 9. Steptoe, A., Dockray, S., & Wardle, J. (2009). Positive affect and psychobiological processes relevant to health. Journal of personality , 77 (6), 1747–1776. Van den Broek, E. L. (2013). Ubiquitous emotion-aware computing. Personal and Ubiquitous Computing , 17 (1), 53–67. Vins, H., Bell, J., Saha, S., & Hess, J. J. (2015). The mental health outcomes of drought: a systematic review and causal process diagram. International journal of environmental research and public health , 12 (10), 13251–13275. Wang Enxu, Z., Jiang, Y., Jun, W., Yicheng, Y., & Pinyan,&Wang Xuejing. (2024). Research on the Impact Mechanism of Built Environment on Urban Vitality Spatial Differentiation at Street Scale Based on MGWR: A Case Study of Shenyang Central Urban Area Geographic Science, 44 (8), 1322–1331. Woo, J., Tang, N., Suen, E., et al. (2009). Green space, psychological restoration, and telomere length. The Lancet , 373 , 299–300. Xie, Bo, Wang, & Xiao,&Wu Lei. (2021). Research on the Impact of Urban Green Space on Residents' Mental Health Based on Natural Experiments: A Case Study of Wuhan Donghu Greenway Progress in Geographic Sciences, 40 (7), 13. Xu, & Dingyuan (2021). Li Xun Parking Problems, Trends, and Countermeasures in Existing Urban Residential Areas [J] (Vol. 28, pp. 25–28). Urban Development Research. 06. Xu, J. (2013). Research on prediction methods based on online user sentiment analysis. Journal of China Library Science , (3), 96–107. Yang Jun, L., & Jing (2013). Lang Senyang Subjective assessment of fatigue status and emotional state of foreign enterprise personnel before and after a single workday [J]. Chinese Journal of Health Psychology , 21 (05), 683–685. Yuen, H. K., & Jenkins, G. R. (2020). Factors associated with changes in subjective well-being immediately after urban park visit. International journal of environmental health research , 30 (2), 134–145. Želinský, T., Hudec, O., Mojsejová, A., & Hricová, S. (2021). The effects of population density on subjective well-being: A case-study of Slovakia. Socio-Economic Planning Sciences , 78 , 101061. Zhang, L. (2023). Research on Parking Problems and Countermeasures of Vehicles in Old Urban Residential Areas Urban Architecture (24), 136–138. Zhang, S. T., Wang, F. F., Duo, F., & Zhang, J. L. (2018). Research on the majority decision algorithm based on WeChat sentiment classification. Journal of Intelligent & Fuzzy Systems , 35 (3), 2975–2984. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-7774299","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":534020922,"identity":"9e5078d2-95a8-4374-ab16-d5bf6be70d7c","order_by":0,"name":"meng Cai","email":"","orcid":"","institution":"Xi'an Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"meng","middleName":"","lastName":"Cai","suffix":""},{"id":534020923,"identity":"e8903b4d-87d9-440a-83f2-474f36e05e87","order_by":1,"name":"xiaoyin Zhang","email":"","orcid":"","institution":"Xi'an Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"xiaoyin","middleName":"","lastName":"Zhang","suffix":""},{"id":534020924,"identity":"aac8ef07-2d91-4f7b-a4b5-e028898c2bee","order_by":2,"name":"xue Gong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA50lEQVRIie3RMYvCMBTA8ZTAcwm6RpT4FSJCP09CIVMPHDs49PBohtM9ouBX0M3NlkKm3O6o38BuTp7uHm3dbshvfn94j4eQ5/1DMCr0rboDCbZlcRbJrDnpUpz3DXQZNhDxs7PNCaMgBgTYpGNI2L984RaLDRZ8/JyWy3WmEpkC6ulvUZ8MHRdXquRqU9qTPAwRdT+7+gTFvDDcyn2q1Ek6QJx+NCbjORG/8pjH4VRmuEVCVYRJDpPAxCFql5DSBqsUWLCwERXOksZbRvozQ1X6fKWeF9UtmbGeXtYnL8h7457ned6fHru3TEm0Jp5jAAAAAElFTkSuQmCC","orcid":"","institution":"Xi'an Jiaotong University","correspondingAuthor":true,"prefix":"","firstName":"xue","middleName":"","lastName":"Gong","suffix":""}],"badges":[],"createdAt":"2025-10-03 14:38:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7774299/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7774299/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94455970,"identity":"83c2b672-1d24-4567-9628-d827b36e2956","added_by":"auto","created_at":"2025-10-27 14:44:04","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":957742,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/a5fb68fc75b3742903d97c40.docx"},{"id":94456261,"identity":"de735736-af94-46c1-ad4d-05aa03a33dde","added_by":"auto","created_at":"2025-10-27 14:44:22","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5792,"visible":true,"origin":"","legend":"","description":"","filename":"12786ef2135b4477aa35b843920f183e.json","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/93b5224cc41b131c406d30f1.json"},{"id":94456315,"identity":"ff941c9d-6925-4ea0-84b1-e90ea71b5ff8","added_by":"auto","created_at":"2025-10-27 14:44:25","extension":"doc","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12800,"visible":true,"origin":"","legend":"","description":"","filename":"TitlePagewithAuthorDetails.doc","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/241bf412ea7cc2de68a24ebb.doc"},{"id":94456159,"identity":"97e7b1d9-1675-4fea-8d3f-92b858d1f5d9","added_by":"auto","created_at":"2025-10-27 14:44:17","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":155285,"visible":true,"origin":"","legend":"","description":"","filename":"12786ef2135b4477aa35b843920f183e1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/a060caecedb9564d1ed173c4.xml"},{"id":94456151,"identity":"392f7d6b-de9f-4157-818a-062000b50571","added_by":"auto","created_at":"2025-10-27 14:44:15","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120796,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/d8165e7185288e73dfe26e0e.png"},{"id":94456434,"identity":"610e4f91-8cb4-4dbf-ad17-962b6e61989d","added_by":"auto","created_at":"2025-10-27 14:44:29","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":80981,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/933b11c4998932abec9fbe8b.png"},{"id":94455545,"identity":"a73b6c51-d692-4d34-b536-091639a50ff7","added_by":"auto","created_at":"2025-10-27 14:43:47","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":146509,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/9cdc7a8690564f77224e1034.png"},{"id":94455985,"identity":"2ad38cad-ce82-47ce-9d0f-1c803f9ba848","added_by":"auto","created_at":"2025-10-27 14:44:05","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":189309,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/def28ee06cc9b9e9a96f1e3c.png"},{"id":94456123,"identity":"6c1724f7-0a52-496d-ba1c-e2755a2136b9","added_by":"auto","created_at":"2025-10-27 14:44:11","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13718,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/bc9b8394da3b4010d542b1d1.png"},{"id":94456281,"identity":"c36358de-bad8-4d36-a75d-6bd96b1ba2c6","added_by":"auto","created_at":"2025-10-27 14:44:23","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9505,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/8f7ecba6e1c1aa197f06ebde.png"},{"id":94455450,"identity":"c6443f91-4631-42f4-84e6-073b403116e6","added_by":"auto","created_at":"2025-10-27 14:43:44","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52848,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/b7d9957aaf85f108a6f62987.png"},{"id":94456298,"identity":"23a133d4-326d-4a3f-911e-ec158c7ab6e0","added_by":"auto","created_at":"2025-10-27 14:44:23","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27378,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/108f0c212d132c9d4a50e526.png"},{"id":94456595,"identity":"78120213-0c4d-4f71-9d3c-74e591647767","added_by":"auto","created_at":"2025-10-27 14:44:45","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68206,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/1c32f5d2be40f77aa2f5f60b.png"},{"id":94456259,"identity":"dc863fcb-8c01-4beb-ada8-ebd9d8efcd82","added_by":"auto","created_at":"2025-10-27 14:44:22","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":62916,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/268455040abfc1b045bd1a96.png"},{"id":94456260,"identity":"1394daf6-38c1-44a2-a9e1-9818d7882e8c","added_by":"auto","created_at":"2025-10-27 14:44:22","extension":"xml","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":147839,"visible":true,"origin":"","legend":"","description":"","filename":"12786ef2135b4477aa35b843920f183e1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/bd4c815dd8682b8bd44ae52d.xml"},{"id":94455564,"identity":"b7060e49-b7f4-4724-bf9f-d660a8d32183","added_by":"auto","created_at":"2025-10-27 14:43:48","extension":"html","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":168397,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/edb5037815655eac3a4bef6f.html"},{"id":94456146,"identity":"9e08de95-119e-4e01-bde9-9f62d1136d72","added_by":"auto","created_at":"2025-10-27 14:44:14","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":150923,"visible":true,"origin":"","legend":"\u003cp\u003eStudy Area Map\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/194274d6e63159ee8dee26a1.jpeg"},{"id":94456596,"identity":"1390ea1a-0c9b-4dc7-90b9-b37c62834f4c","added_by":"auto","created_at":"2025-10-27 14:44:45","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":191250,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of different built environments in the study area\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/403bf843ae7d4be368bea10b.jpeg"},{"id":94455491,"identity":"da0a0847-dc2e-4bd4-9b4a-40a85bc38c86","added_by":"auto","created_at":"2025-10-27 14:43:46","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":123511,"visible":true,"origin":"","legend":"\u003cp\u003ea month scale change chart of emotion value\u003c/p\u003e\n\u003cp\u003eb weekly scale change chart of emotion value\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/42c06e428220648d6dfc09e3.png"},{"id":94455456,"identity":"2c9b6322-63fb-4abb-8c9b-01639f570d8d","added_by":"auto","created_at":"2025-10-27 14:43:44","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":65716,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of residents' emotions in the study area\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/cb596f653b22fc969b220047.jpeg"},{"id":94454405,"identity":"96f903f4-5629-4ff4-b4ab-179ce3395220","added_by":"auto","created_at":"2025-10-27 14:43:28","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":92873,"visible":true,"origin":"","legend":"\u003cp\u003ea Analysis of the importance of characteristic variables\u003c/p\u003e\n\u003cp\u003eb SHAP feature summary plot\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/3d93c176ce261717fc6e76ec.jpeg"},{"id":94454474,"identity":"ff8fb15b-6f4c-44e2-8a38-1cbc4e86e812","added_by":"auto","created_at":"2025-10-27 14:43:29","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":124691,"visible":true,"origin":"","legend":"\u003cp\u003eMarginal effects of feature variables on residents' emotional health\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/16f563def5c1425ea58f93d7.jpeg"},{"id":94455453,"identity":"f63e9194-d73a-43a5-bbe3-695beef47523","added_by":"auto","created_at":"2025-10-27 14:43:44","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":99048,"visible":true,"origin":"","legend":"\u003cp\u003eThe local interaction effect of built environment variables on residents' emotional health\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/b4fc556358a64505fdbf84f5.jpeg"},{"id":101753197,"identity":"3ee634eb-1369-437d-886f-b28bd961b4cf","added_by":"auto","created_at":"2026-02-03 10:39:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1795060,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7774299/v1/d8a19fc4-2c49-4a88-a5d1-2a8f7abaaf97.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Decoding Urban Emotions: Exploring the Association between Urban Built Environment and Residents' Emotional Health Using Interpretable Machine Learning Models","fulltext":[{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e\u003ch2\u003e1.1 Research background\u003c/h2\u003e\u003cp\u003eWith the continuous advancement of global urbanization, it is predicted that by 2050, nearly 70% of the global population will reside in urban areas (Heilig, 2012). Cities, as the primary spaces for population aggregation, serve not only as engines of economic growth but also as hubs for daily life, leisure, and entertainment. However, the \"big city disease\" and unhealthy lifestyles resulting from rapid urbanization have imposed a substantial burden on residents' physical and mental health (Gong et al., 2012). Among these issues, mental health problems are more likely to be overlooked due to their uncertainty and intangibility. Sustainable cities and communities (SDG11) have been identified by the United Nations as one of the key agendas for 2030, emphasizing the improvement of the built and social environments in urban planning and community development to enhance residents' health and well-being. In China, the concept of a \"healthy city\" has also emerged as one of the significant goals to be achieved by 2030. This goal necessitates the integration of health considerations into every aspect of urban planning, construction, and management, thereby promoting residents' health and bolstering urban sustainability.\u003c/p\u003e\u003cp\u003eEmotion is a product of the interaction between individuals and their environment. Psychologist Broek (2013) identified six basic emotional categories: anger, disgust, fear, happiness, sadness, and surprise. From these fundamental emotions, other non-basic emotions can be derived, such as relaxation, oppression, anxiety, and security. While emotional research is prevalent in psychology, sociology, and medicine, the growing interdisciplinary exchanges have led geography, architecture, and related fields to increasingly focus on the study of emotion and space. Emotional health denotes a stable and positive emotional state that confers health benefits (Guo \u0026amp; Du, 2023). Existing research indicates that emotions are closely linked to individuals' physical health, mental health, and social adaptation. Particularly, positive emotions can reduce the risk and likelihood of illness, enhance physical and mental well-being, and improve quality of life by influencing cognitive processes and encouraging healthier behaviors (Steptoe et al., 2009; Steptoe \u0026amp; Diez, 2008). Moreover, positive emotions are associated with broader social connections and support. For instance, positive emotions can expand individuals' interpersonal resources, such as friendships and social networks, which in turn provide social support and mutual aid, helping individuals effectively navigate various life challenges and difficulties (Guo \u0026amp; Wang, 2007).\u003c/p\u003e\u003cp\u003eThe urban built environment mainly refers to the physical spaces such as urban functional spaces and transportation systems constructed through human planning, design, construction and renovation, including housing, schools, commercial areas, roads, etc. (Handye et al., 2005). Initial research on the built environment identified evaluation indicators across three dimensions\u0026mdash;density, diversity, and design\u0026mdash;establishing a 3D framework for assessing the built environment (Cervero \u0026amp; Kockelman, 1997). Subsequently, two additional dimensions\u0026mdash;destination accessibility and distance to transportation hubs\u0026mdash;were incorporated into this framework, resulting in a more comprehensive 5D indicator system for the built environment. This system, which includes density, diversity, design, destination accessibility, and distance to transportation stations (Ewing et al., 2010), has become the principal basis for contemporary evaluations of the built environment. The built environment not only defines the daily living spaces of residents but also significantly influences their emotional states. Specifically, the walkability of the built environment directly impacts residents' daily activities and emotions. A walkable community environment can markedly enhance physical activity levels, thereby boosting positive emotions and reducing stress (Hematian \u0026amp; Ranjbar, 2022). Moreover, accessibility to service facilities and entertainment venues is crucial, as these spaces provide essential social and leisure areas that help alleviate stress and increase happiness, thus positively affecting residents' emotions (S\u0026auml;umel et al., 2016). Additionally, a well-designed community fosters neighborly communication and cooperation, enhances community cohesion, and instills a greater sense of safety and belonging among residents. Conversely, issues such as poor walking experiences, walking fears, or spatial disorganization can heighten anxiety, diminish happiness, and even deter healthy behaviors (Aram et al., 2019; Li et al., 2015).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e1.2 Literature review\u003c/h2\u003e\u003cp\u003eResearch on the urban built environment and emotional health can be traced back to the concept of emotional geography proposed by British scholars Anderson and Smith in 2001. They formally introduced the concept of emotional geography, exploring the interaction between human emotions\u0026mdash;such as joy, anger, sorrow, and happiness\u0026mdash;and various spatial environments. This marked a shift in the study of human perception and the urban environment from purely subjective psychological domains to a broader social spatial dimension (Anderson \u0026amp; Smith, 2001). Currently, with the improvement of urban residents' quality of life and the transformation of urban construction from a stage of incremental expansion to one focusing on quality enhancement, the relationship between human behavior and the built environment has garnered increased attention. The connection between emotional activities and urban space has also become a key research focus. Numerous previous studies have explored the interaction between urban space and public sentiment, yielding meaningful results (He et al., 2024). For example, Chen et al. (2018) selected three streets with distinct characteristics in Beijing as research subjects to investigate the correlation between the landscape features of different streets and people's emotional responses. Han et al. (2024) compared the impact of urban-scale, community-scale, and residential-scale built environments on the mental health of the elderly, using Hefei City in Anhui Province, China, as a case study, and found that the urban-scale built environment had the highest relative importance. Chen et al. (2016) utilized emotion mapping analysis technology, combined with biofeedback instruments and GPS, to collect physiological indicators and geographical data from specific populations during real-life experiences. They studied the emotional responses elicited by the real-life environment and further explored the use of fractal measurement technology to construct an accurate emotion measurement model.\u003c/p\u003e\u003cp\u003ePrevious studies have shown that urban morphology, landscape environment and entertainment facilities have a significant impact on public sentiment (Chen et al. 2018; He et al., 2024). For example, researchers observed that adequate and high-quality urban greening and blue spaces such as lakes and rivers help trigger positive emotions (Huai \u0026amp; Van, 2022), in contrast, adverse environmental conditions such as noise, congestion and heat island effects, can lead to negative emotions (Holly et al., 2015). However, the researchers found that the relationship between urban environment and public sentiment was not comprehensively consistent. For example, many studies have shown that road networks bring many conveniences to people's lives and travel, but Perez et al. (2022) observed that excessive road density will bring negative effects such as noise disturbances, environmental pollution and traffic congestion. Similarly, many studies have reported that high building density tends to cause negative emotions (He et al., 2022), however, Huang et al. observes that high building density in Hong Kong can also promote positive emotions among residents (Huang et al., 2023), because There may be a synergy between building density and other factors such as infrastructure quality and amenities. It can be seen from this that the impact of urban spatial environment on residents' lives and emotions is not always a linear and single direct relationship, but may have opposite effects or mixed effects with other factors after exceeding a certain threshold. Therefore, it is necessary to deeply analyze whether there is a complex relationship between the built environment and emotional health.\u003c/p\u003e\u003cp\u003eIn addition, existing studies often describe residents' emotional characteristics at macro-levels such as cities, streets, and communities, exploring the relationship between individual emotions and spatial environments (Liu et al., 2020; Chen et al., 2018). In contrast, research on the spatial and temporal distribution characteristics of average emotions within grid units and their relationship with the built environment, particularly at the micro-scale of 15-minute living circles, is still in its exploratory phase. The 15-minute city living circle is a people-oriented urban planning concept based on time urbanism, envisioning that residents can easily access basic services and amenities, such as convenience stores, healthcare facilities, schools, and parks, within a 15-minute walk or cycle from their homes (Murgante et al., 2024; Moreno et al., 2021). As the global urbanization process deepens and people's demands for living environments and quality of life increase, the 15-minute city living circle concept is gaining widespread attention. An increasing number of cities worldwide are integrating this concept into urban planning and construction. For example, the C40 City Initiative, involving 100 major cities across multiple countries, emphasizes the importance of implementing 15-minute cities. Cities such as Milan, Paris, Bogota, Portland, and Melbourne have taken actions to reconfigure the built environment to promote intelligent, low-carbon, and sustainable community development (C40 Cities Climate Leadership Group, 2020; Allam et al., 2022). Since China introduced the 15-minute city living circle planning guide in Shanghai, many Chinese cities, including Jinan, Guangzhou, Hangzhou, and Suzhou, have incorporated this concept into their planning documents and gradually promoted it in national urban renewal efforts (Nie, 2024). Therefore, it is essential to deepen the discussion of the relationship between the built environment and residents' health at the micro-geographical level of the 15-minute city living circle.\u003c/p\u003e\u003cp\u003eFurthermore, current research often focuses on describing residents' emotional characteristics at the macro level, such as cities, streets, and communities, exploring the relationship between individual emotions and the spatial environment (Liu et al., 2020; Chen et al., 2018). In contrast, research examining the spatiotemporal distribution characteristics of residents' emotions within micro-grid units, especially from the perspective of living circles, and their nonlinear relationships, threshold effects, and interactive effects with the built environment, is still in the exploratory phase. The concept of living circles is a people-oriented urban planning philosophy, based on the concept of temporal urbanism, envisioning that residents can easily access basic services and conveniences, such as convenience stores, healthcare facilities, schools, and parks, within approximately 5, 10, or 15 minutes' walk or bike ride from their residential areas (Murgante et al., 2024; Moreno et al., 2021). With the deepening of the global urbanization process and the increasing demands for living environments and quality of life, the concept of living circles is gradually receiving widespread attention. An increasing number of cities worldwide are incorporating the living circle philosophy into urban planning and construction, such as the C40 Cities Initiative, which involves 100 major cities from several countries, emphasizing the importance of implementing the 15-minute city, and cities like Milan, Paris, Bogot\u0026aacute;, Portland, and Melbourne, have already implemented actions focused on reconfiguring the built environment to promote intelligent, low-carbon, and sustainable community development (C40 Cities Climate Leadership Group, 2020; Allam et al. 2022). Since Shanghai introduced the 15-minute CLC planning guidelines, many Chinese cities, including Jinan, Guangzhou, Hangzhou, and Suzhou, have incorporated this concept into their planning documents and are gradually promoting its practice in urban renewal nationwide (Nie, 2024). Therefore, it is necessary to deepen the exploration of the relationship between the built environment and residents' health at the micro geographic level of living circles.\u003c/p\u003e\u003cp\u003eFinally, existing research primarily relies on questionnaires and interviews to gather data. However, due to limitations in scale and spatial and temporal precision, these methods struggle to fully capture the relationship between large areas, large samples, and the real-time comprehensive urban built environment and residents' emotions. With the rise of social media applications, people have increasingly become accustomed to sharing their daily lives and expressing their opinions on these platforms. Against this backdrop, location-based social network (LBSN) data, such as that from Weibo and Twitter, has emerged and rapidly provided a wealth of real-time public sentiment data. These emerging social media big data sources offer potential solutions to the shortcomings of traditional public opinion research methods (Shan et al., 2022). Despite ongoing controversies regarding privacy, representation, and other issues (Murthye et al., 2015), social media data is progressively supplanting traditional survey data as \u0026ldquo;social sensors\u0026rdquo; to enhance our understanding of real-world social phenomena.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e1.3 Research content and contributions\u003c/h2\u003e\u003cp\u003eTo summarize, This paper first collects user posting information on Weibo social media in Nanning City, Guangxi, from January to December 2023, and employs the SnowNLP library in Python to analyze the emotion values of the cleaned posts. It also establishes a 15-minute city living circle scope to create independent grids, calculates the average emotional values of residents in each grid, and analyzes their spatial and temporal distribution characteristics. Secondly, this paper constructs a 5D environmental index system, including key indicators such as population density, building density, functional mix, road intersection density, road network density, greening rate, commercial service centers, and public transportation station density. Based on this, the gradient decision tree (XGBoost) machine learning method is used to identify the contribution ranking and characteristic effects of different built environmental factors on residents' emotional health. Finally, the paper discusses the complex nonlinear relationships and threshold effects between various built environmental factors and residents' emotional health.\u003c/p\u003e\u003cp\u003eThe contributions of This paper are as follows: First, by constructing a grid unit based on the 15-minute living circle, this study can more precisely capture and analyze the distribution characteristics of residents' emotions in small-scale spaces and examine the subtle effects of the built environment within the 15-minute living circle on residents' emotional health. Second, this paper utilizes social media big data and built environment big data to expand the data sources, enhance the timeliness and comprehensiveness of the research, and provide a more accurate and real-time perspective for evaluating and understanding the impact of the urban built environment on residents' emotional health. This, in turn, offers robust data support and a scientific basis for the development of urban planning and public health interventions. Third, based on the XGBoost machine learning model and the SHAP interpretability framework, we explore the nonlinear relationships and threshold effects between various built environmental elements and residents' emotions, thereby compensating for the limitations of traditional regression methods that rely solely on linear analysis.\u003c/p\u003e\u003c/div\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Research area\u003c/h2\u003e\u003cp\u003eThe research area of This paper primarily focuses on the central urban area of Nanning City, the capital of Guangxi Autonomous Region, located in western China. There are two main reasons for selecting Nanning City. Firstly, as the capital and economic and cultural hub of Guangxi Zhuang Autonomous Region, Nanning City exhibits a high degree of urbanization and population concentration. Its urban built environment and residents' living conditions are highly representative. Moreover, the significant population aggregation results in abundant and accessible social media data (such as Weibo), which can provide a wealth of residents' emotional health data for research purposes. Secondly, most prior studies have concentrated on developed cities, with fewer investigations into underdeveloped cities in less developed regions. As a city in the underdeveloped western part of China, Nanning can offer valuable insights and data support for other similar regions through this research, assisting urban planners and policymakers in better incorporating residents' emotional health into urban construction, optimizing the layout of the built environment, and enhancing residents' quality of life. Nanning City comprises 7 urban districts, 4 counties, and 1 county-level city. Among these, Qingxiu District, Xixiangtang District, Jiangnan District, Xingning District, Liangqing District, and Yongning District are the six main urban districts (Nanning Municipal Government, 2023), with the central urban area being the confluence of these districts. The research area was delineated based on the urban central street, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. According to the \"Standards for Planning and Design of Urban Residential Areas,\" the walking distance for a 15-minute living circle should be 800\u0026ndash;1100 meters to meet basic living needs (Ministry of Housing and Urban-Rural Development, 2018). Therefore, this study adopted an 800m\u0026times;800m grid unit, independently dividing the research area into grids of this size. Subsequent statistics on built environmental factors, residents' emotional health, and corresponding data analysis were all conducted based on these grid units.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Research data\u003c/h2\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003e2.2.1 Urban built environment data\u003c/h2\u003e\u003cp\u003eThe 5D built environment indicator system encompasses five dimensions: density (including population and building density), diversity (pertaining to the variety of urban functional land uses and activity types), design (the planning and design of blocks and buildings), distance (to transportation facilities and service centers), and accessibility (including the convenience of walking, cycling, and public transportation) (Ewing et al., 2010; Wang Enxu et al., 2024). Accordingly, This paper adopts the established 5D built environment indicator system to assess the built environment level of Nanning's central urban area. Drawing on existing research, the characteristics of Nanning's central urban area, and data availability, This paper selects nine indicators: population density, building density, mixed use of urban functional land, road network density, greening rate, commercial service centrality, indoor sports and leisure service centrality, and density of public and transportation stations in parks and green spaces (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Using ArcGIS 10.8, we calculated and extracted the built environment variables within each 800m\u0026times;800m grid cell based on the selected indicators and built environment data. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates the spatial distribution of the nine types of built environment indicators, revealing significant spatial differentiation patterns across different levels of the built environment.\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\u003e5D built environment indicators and sources\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\" colname=\"c1\"\u003e\u003cp\u003eDimension\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eindex\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCalculation method\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSource\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003edensity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePopulation density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePopulation number/grid area (people/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eChina's 7th Census Data\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.stats.gov.cn/sj/z\u003c/span\u003e\u003cspan address=\"https://www.stats.gov.cn/sj/z\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003cp\u003exfb/202302/t20230203_1901080.html?\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eBuilding density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBuilding base area/grid area (m)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLandsat8 satellite imagery\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://earthexplorer.usgs.gov/\u003c/span\u003e\u003cspan address=\"https://earthexplorer.usgs.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ediversity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUrban functional land mix\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eArea ratio of various functional land in the grid\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eOpenStreetMap\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openmaptiles.org/\u003c/span\u003e\u003cspan address=\"https://openmaptiles.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003edesign\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRoad network density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTotal length of the road network in the grid/grid area(m2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eOpenStreetMap\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openmaptiles.org/\u003c/span\u003e\u003cspan address=\"https://openmaptiles.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGreening rate\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNormalized vegetation index NDVI mean in grid\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGeospatial Data Cloud\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gscloud.cn/\u003c/span\u003e\u003cspan address=\"https://www.gscloud.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eDestination accessibility\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eBusiness Service Center\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eThe average core density of commercial service POI in the grid, such as clothing stores, supermarkets and restaurants, etc.\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGaode Map Open Platform\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://developer.amap.com/\u003c/span\u003e\u003cspan address=\"https://developer.amap.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003endoor sports and leisure service center\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMean core density of POI in the grid sports and leisure service category, such as gyms and cinemas\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGaode Map Open Platform\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://developer.amap.com/\u003c/span\u003e\u003cspan address=\"https://developer.amap.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePark green space\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eThe mean core density of scenic spots and services in the grid, such as parks, green squares, etc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGaode Map Open Platform\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://developer.amap.com/\u003c/span\u003e\u003cspan address=\"https://developer.amap.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTraffic stop distance\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTraffic site density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eThe mean core density of traffic stations in the grid, including three categories: bus, subway and private car parking stations\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGaode Map Open Platform\u003c/p\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://developer.amap.com/\u003c/span\u003e\u003cspan address=\"https://developer.amap.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\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\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\u003ch2\u003e2.2.2 Resident emotional health data\u003c/h2\u003e\u003cp\u003eThe data utilized in This paper originates from Sina Weibo, one of China's largest social media platforms. In 2023, Weibo boasted 598\u0026nbsp;million users, akin to Twitter, which is globally widespread. It permits users to post text, images, videos, and links to document and share their lives, express emotions, and voice opinions on various events and issues. This study amassed 219,960 Weibo posts related to Nanning city, geotagged with location markers, published on the Sina Weibo platform from January to December 2023. Each post encompasses content, release time, location, authentication type, and author details. The dataset underwent desensitization, noise reduction, and removal of invalid samples from non-study areas, culminating in 188,103 Weibo posts. Subsequently, SnowNLP was employed to analyze and process the text content of these posts. SnowNLP is a Python-based method specifically designed for processing Chinese text, capable of accurately calculating the probability of a post's sentiment being positive, with an accuracy rate exceeding 80% (Zhang et al., 2018). Specifically, texts conveying joy, happiness, satisfaction, or admiration are categorized as \"positive\" emotions, whereas those expressing sadness, anger, disappointment, or depression are deemed \"negative\" emotions. The emotional probability value ranges from 0 to 1. Drawing on existing research and aligning with the content and data characteristics of this study, an emotional score of 0.6 and below is considered a low mood value, 0.6 to 0.8 is a medium mood value, and above 0.8 is a high value (Shan et al., 2022). Lastly, following He et al. (2024), emotional health value is defined as the average of emotional values from all Weibo users' posts within any 800m\u0026times;800m grid cell to reflect the emotional health of residents in that area. Grids without any Weibo user posts are treated as missing values and excluded from subsequent data analysis.Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the descriptive statistics of each variable.\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\u003eDescriptive statistics of variables\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" 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\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSD\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMin\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMax\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eemotional health value\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.730\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.205\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epopulation density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.006\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0..007\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.034\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ebuilding density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.135\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.093\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.447\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eland use mix\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.012\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.010\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.030\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eroad network density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.008\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.005\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.031\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNDVI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.333\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.123\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.009\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.766\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ecommercial service centrality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e253.258\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e295.753\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2623.23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eindoor recreational centrality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.137\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e13.679\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e102.031\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003egreen space centrality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.775\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.425\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e24.533\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003etransit station density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e30.663\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e34.278\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e212.566\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Model method\u003c/h2\u003e\u003cp\u003eThis paper employs nine types of built environment elements as characteristic variables and residents' emotional health as the response variable to construct an interpretable model integrating XGBoost and SHAP to assess the impact on residents' emotional health. The XGBoost model, proposed by Chen et al. (2016), is an efficient algorithm that significantly improves upon the gradient boosting decision tree (GBDT). XGBoost is an ensemble machine learning model based on the gradient boosting algorithm, comprising multiple classification and regression trees (CARTs). Its fundamental principle involves continuously iterating to generate new decision trees that fit the residuals from the preceding tree model and summing the predicted values of these decision trees. Thus, the XGBoost model can be represented as the sum of the scores of k-CARTs, as illustrated below:\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\:\\widehat{{\\mathbf{y}}_{\\mathbf{i}}}={\\sum\\:}_{\\mathbf{k}=1}^{\\mathbf{k}}{\\mathbf{f}}_{\\mathbf{k}}\\left({\\mathbf{x}}_{\\mathbf{i}}\\right)\\:{(\\mathbf{f}}_{\\mathbf{k}}\\in\\:\\mathbf{F})$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eIn the formula (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\widehat{{\\mathbf{y}}_{\\mathbf{i}}}\\)\u003c/span\u003e\u003c/span\u003e is the model predicted value of the i-th sample;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{k}\\)\u003c/span\u003e\u003c/span\u003e is number of trees;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\mathbf{x}}_{\\mathbf{i}}\\)\u003c/span\u003e\u003c/span\u003e is the eigenvector of the i-th sample;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\mathbf{F}\\:\\)\u003c/span\u003e\u003c/span\u003eis a collection of all CART trees,\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\mathbf{f}}_{\\mathbf{k}}\\)\u003c/span\u003e\u003c/span\u003e is the kth CART tree, that is, the predicted value of the x-th sample。\u003c/p\u003e\u003cp\u003eThe overall idea of XGBoost is to combine the loss function and regularization term into a global loss function, and perform second-order Taylor expansion. Its objective function is:\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$$\\:{\\mathbf{S}}_{\\mathbf{o}\\mathbf{b}\\mathbf{j}}\\left(\\varvec{\\theta\\:}\\right)={\\sum\\:}_{\\mathbf{i}=1}^{\\mathbf{N}}\\mathbf{L}({\\mathbf{y}}_{\\mathbf{i}},\\widehat{{\\mathbf{y}}_{\\mathbf{i}}})+{\\sum\\:}_{\\mathbf{j}=1}^{\\mathbf{t}}\\varvec{\\Omega\\:}\\mathbf{\u0026fnof;}\\left(\\mathbf{j}\\right)$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eIn the formula (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e)\u003c/strong\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\sum\\:}_{\\text{i}=1}^{\\text{N}}\\text{L}({\\text{y}}_{\\text{i}},\\widehat{{\\text{y}}_{\\text{i}}})\\)\u003c/span\u003e\u003c/span\u003erepresent the overall loss, express the model error, which is the difference between the true value of the sample and the predicted value;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\sum\\:}_{\\text{j}=1}^{\\text{t}}{\\Omega\\:}\\text{\u0026fnof;}\\left(\\text{j}\\right)\\)\u003c/span\u003e\u003c/span\u003eas a regularization term, it expresses the structural error of the model, i.e. the complexity of the regression tree, by using hyperparameters multiplied by the number and value of nodes to limit the complexity of the model.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\n$$\\:{\\mathbf{S}}_{\\mathbf{w}\\mathbf{h}\\mathbf{e}\\mathbf{r}\\mathbf{e}}\\varvec{\\Omega\\:}\\left(\\mathbf{\u0026fnof;}\\right)=\\varvec{\\gamma\\:}\\mathbf{T}+\\frac{1}{2}\\varvec{\\lambda\\:}{‖\\varvec{\\omega\\:}‖}^{2}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eIn the formula(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e)\u003c/strong\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{\\Omega\\:}\\left(\\mathbf{\u0026fnof;}\\right)\\)\u003c/span\u003e\u003c/span\u003eis a regularization term of the model, used to reduce the overfitting problem and complexity of the model;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\mathbf{T}\\)\u003c/span\u003e\u003c/span\u003eis the number of leaf nodes;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{\\lambda\\:}\\)\u003c/span\u003e\u003c/span\u003efor the severity of punishment;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{\\omega\\:}\\)\u003c/span\u003e\u003c/span\u003e is output scores for leaf nodes;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\frac{1}{2}\\varvec{\\lambda\\:}{‖\\varvec{\\omega\\:}‖}^{2}\\)\u003c/span\u003e\u003c/span\u003e is the L2 modulus square of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{\\omega\\:}\\)\u003c/span\u003e\u003c/span\u003e.If the objective function of the model is smaller, the better the prediction effect of the model.\u003c/p\u003e\u003c/p\u003e\u003cp\u003eThe XGBoost model offers numerous advantages. Firstly, unlike traditional linear regression models, XGBoost does not presuppose a specific relationship between independent and dependent variables. Instead, it learns and automatically captures the complex nonlinear relationships by constructing decision trees, which traditional linear regression models cannot achieve. Secondly, compared to other tree-based models (such as GBDT), XGBoost is less sensitive to multicollinearity, missing values, outliers, and irrelevant variables. It can handle high-dimensional data effectively, is less prone to overfitting, and is highly adaptable to various datasets. It can process both discrete and continuous data without the need for normalization.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Result","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.1 The spatiotemporal distribution characteristics of residents' emotions in the research area\u003c/h2\u003e\u003cdiv id=\"Sec13\" class=\"Section3\"\u003e\u003ch2\u003e3.1.1 Time-change characteristics\u003c/h2\u003e\u003cp\u003eBased on the emotion values of all Weibo users posted within the research area, This paper first calculates the mean and median emotion values of Weibo users' posts on a monthly and weekly scale. Figure\u0026nbsp;3a illustrates the fluctuation trend of Weibo user sentiment in the study area throughout the year, with the median consistently above the mean and exhibiting similar fluctuation patterns. Specifically, the period from the end of the year to the beginning of the year, particularly January, marks a peak in user emotions, likely due to the festive atmosphere of holidays such as New Year's Day and the Chinese Spring Festival. During these holidays, people use social media platforms like Weibo to record and share joyful moments and New Year's greetings, thereby enhancing emotional value. In February, sentiment values dropped, with the mean and median falling to 0.731 and 0.843, respectively. This decline may be related to the adjustment of life routines following the Chinese Spring Festival holiday. The end of the holiday signifies the start of long-term work commitments and the separation from family, friends, and hometown, as well as the reluctance to leave the pleasant holiday times, all of which contribute to a decrease in residents' emotional values. From March to October, emotions remained relatively stable, with the mean fluctuating slightly between 0.740 and 0.752. After October, sentiment values rose again, with the mean and median in December reaching 0.799 and 0.914, respectively, marking another annual high.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;3b depicts the weekly emotional fluctuations of residents. The data indicates that emotional values are higher on rest days compared to weekdays, with Sunday recording the highest sentiment scores. This trend may be attributed to the \"weekend effect.\" Following a week of work or study-related stress, individuals typically engage in more restful and relaxing activities over the weekend, such as family dinners, outdoor pursuits, or personal hobbies. These leisurely experiences render the weekend the most emotionally positive period of the week (Liu \u0026amp; Chai, 2001; Shan et al., 2022). Starting from Monday, emotion values gradually decline, reaching their nadir on Thursday. The particularly low emotional values observed on Thursday could be linked to the \"midweek slump.\" People might feel somewhat fatigued by Thursday, having worked through most of the week without the imminent prospect of weekend relaxation and entertainment. This \"neither here nor there\" phase may temporarily dampen spirits (Yang et al., 2013). However, as the weekend approaches,emotion begins to recover on Friday and peaks once more on Saturday.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e\u003ccolgroup cols=\"1\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFigure\u0026nbsp;3a month scale change chart of emotion value\u003c/p\u003e\u003cp\u003e Fig.\u0026nbsp;3b weekly scale change chart of emotion value\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\u003ch2\u003e3.1.2 Spatial distribution characteristics\u003c/h2\u003e\u003cp\u003eBased on the magnitude of emotional values and in conjunction with the research by Shan et al. (2022), this paper categorizes the grid units of the research area into three groups: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) Areas with residents' emotional values of 0.8 and above are classified as high-emotion areas; (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) Areas with emotional values between 0.6 and 0.8 are designated as medium-emotion areas; (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) Areas with residents' emotional values below 0.6 are considered low-emotion areas. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e reveals that, generally, the distribution of emotional space in the study area is relatively fragmented, with a mixed distribution of high, medium, and low emotional values. It is noteworthy that the average emotional health value across the entire grid is 0.730, with a standard deviation of 0.205 (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), indicating that the overall mood in Nanning is relatively positive and healthy. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e suggests a potential spatial correlation between urban environmental factors and residents' emotional health. For instance, compared to low and medium-emotion areas, high-emotion areas are more scattered and have a more pronounced distribution on the urban periphery. This may be because, as the city expands, new residential districts and commercial areas are developed outside the city. These areas may emerge as \"new highlands\" for emotional health due to factors such as low population density, less traffic congestion, and better natural environmental quality. This also reflects the \"edge effect\" of urban life: urban residents may feel alienated from crowded urban centers and may be more inclined to seek quieter and more private living spaces on the urban fringe to achieve a better emotional experience. This finding is consistent with a study conducted in San Francisco, USA. Based on sentiment analysis of the city's Twitter data, the study found that the periphery of cities with abundant vegetation resources exhibited significantly higher sentiment indexes than areas with limited vegetation resources (He et al., 2024), suggesting that individuals in areas with specific environmental conditions are more likely to develop positive emotions.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.2 SHAP-based explanatory enalysis of the model\u003c/h2\u003e\u003cp\u003eSubsequently, this study employs the \"xgboost\" library in Python to analyze the intricate relationship between built environment variables and emotional indices. Initially, 80% of the samples are randomly designated as the training set to train the model, while the remaining 20% serve as the test set to assess model performance. Subsequently, the model's generalization ability is evaluated using a 5-fold cross-validation method to mitigate the risk of overfitting. Specifically, the training set is divided into five subsets, with four used for training in each iteration and the remaining subset used for validation. Following this, hyperparameter tuning is conducted using grid search, setting the learning rate to 0.1 (learning_rate: 0.1), selecting 50 decision trees (n_estimators: 50), with each tree having a maximum depth of 3 (max_depth: 3). Additionally, the subsample ratio for each tree is set to 0.7 (subsample: 0.7), and the ratio of randomly selected features is set to 0.8 (colsample_bytree: 0.8). The performance of the XGBoost model is assessed using mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE), with respective values of 0.036, 0.189, and 0.147. These results demonstrate the model's exceptional accuracy and predictive capabilities.\u003c/p\u003e\u003cp\u003eAfter further exploring the complex relationship between the built environment and residents' emotional health using the XGBoost model, this paper employs the SHAP interpretation framework to visualize and explain the results of the XGBoost model. SHAP is a classic post-hoc attribution interpretation framework (Parsa et al., 2020) that not only outputs the importance ranking of variables but also more intuitively represents the local positive or negative effects of all samples of a variable on the model. Additionally, the output partial dependence plot (PDP) results can describe the degree of change of variables across different values, thereby enhancing the interpretability of the model.\u003c/p\u003e\u003cdiv id=\"Sec16\" class=\"Section3\"\u003e\u003ch2\u003e3.2.1 Analysis of the importance of characteristic variables\u003c/h2\u003e\u003cp\u003eThis paper integrates the SHAP interpretation framework to visualize and elucidate the analysis results of the XGBoost model. The SHAP mean value is utilized to illustrate the relative importance of characteristic variables, specifically the influence of various built-environment factors on residents' emotional health (Fig.\u0026nbsp;5a). It is evident that public transportation stations and green spaces are the most critical factors affecting emotional health, exerting a substantial impact on residents' emotions. Additionally, the density of the road network and the centrality of indoor entertainment facilities significantly contribute to emotional health. In contrast, building density, land use mixing, and the normalized vegetation index have relatively minor impacts, aligning mostly with anticipated and existing research findings. Although the Normalized Difference Vegetation Index (NDVI) demonstrated an important role, its impact was lower than expected and reported in relevant studies. This discrepancy may be attributed to Nanning, known as China's \"Greentown,\" situated south of the Tropic of Cancer. With a humid subtropical monsoon climate, abundant sunshine, and plentiful rainfall, Nanning has an average annual temperature of around 21.6 degrees Celsius (China Weather Network, 2023). The vegetation is lush and evergreen year-round, and the urban greening coverage rate reaches 40.90% (Nanning Municipal Government, 2020). Streets, parks, and communities are lined with numerous trees and flowering plants, creating a uniform urban landscape. Consequently, residents accustomed to this pervasive green environment may experience minimal impact on their lives from variations in greening.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section3\"\u003e\u003ch2\u003e3.2.2 Local feature effects of feature variables\u003c/h2\u003e\u003cp\u003eOn the basis of identifying the importance of feature variables, this paper uses SHAP feature summary plot to combine feature importance with feature effects, reflecting the strength, distribution, and direction of the influence of feature variables on predictor variables (Parsa et al., 2020), in order to reveal the complex relationship between various constructed environmental features and emotional health. As shown in Fig.\u0026nbsp;5b, on the vertical axis, each row represents a feature variable, namely the constructed built environment variable, and each point represents a sample. The redder the color of the point, the larger the value of the feature itself, and the bluer the color, the smaller the value of the feature itself. The horizontal axis represents the SHAP value of the feature variable, and the directions on both sides indicate positive or negative effects. If SHAP value\u0026thinsp;\u0026gt;\u0026thinsp;0, it indicates that the feature variable, namely the built environment element, has a positive effect on residents' emotional health; SHAP value\u0026lt;0, This indicates that the feature variable has a negative effect on residents' emotional health |SHAP Value| represents the magnitude of the lift or decrease force on the predictor variable when this feature is used as a condition. For example, taking Land use mix as an example, the red line on the right side of Land use mix is longer than the blue line on the left side, that is, the red | SHAP Value | is greater than the blue, indicating that high land use mix can significantly affect residents' emotional health more than low land use mix. In addition, the Shapley values of high-value samples (red dots) in Land use mix are mostly positive, while the Shapley values of low value samples (blue dots) are mainly negative, indicating that high land use mix can promote residents' emotional health, while low land use mix can inhibit residents' emotional health. Although this graph provides valuable information, it cannot accurately quantify the impact of different values of the built environment on emotional health. Therefore, this study will use Partial Dependency Plot (PDP) to further elucidate the complex impact of built environment on emotional health within different value ranges.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e\u003ccolgroup cols=\"1\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;5a Analysis of the importance of characteristic variables Fig.\u0026nbsp;5b SHAP feature summary plot\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section3\"\u003e\u003ch2\u003e3.2.3 Nonlinear relationship and threshold effect between built environment and residents' emotional health\u003c/h2\u003e\u003cp\u003eThis paper further employs PDP to illustrate the impact of different built-environment characteristics on residents' emotional health under varying value conditions (Parsa et al., 2020). For each of the nine sets of graphs, the horizontal axis represents the feature variable, which is the constructed built-environment variable, and each point represents a sample. The larger the value, the higher the density value of the feature variable. The vertical axis represents the magnitude and direction of the impact of the built-environment variable on residents' emotional health (predicted value). If the impact is positive, it indicates that the feature has enhanced the predicted value; if the impact is negative, it indicates that the feature has diminished the predicted value. The absolute value represents the magnitude of the lifting or lowering force when this feature is used as a condition. From the partial dependency graph, it can be seen that each feature variable has an overall impact on residents' emotional health, as well as the direction and intensity of its specific values. Based on this, the mode of action of this feature variable can be analyzed (Lundberg et al., 2019)..\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFrom the results shown in the dependency graph in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003e, it can be observed that there is a significant nonlinear relationship and threshold effect between the built environment elements and emotional health. Specifically,to be more specific:\u003c/p\u003e\u003cp\u003ePopulation Density: Within the range of 0-0.005, population density has a positive impact on emotional health; beyond 0.005, the influence shifts to negative and stabilizes after 0.01. This suggests that moderate population density may enhance emotional health due to increased economic activities and job opportunities, but excessively high density can have adverse effects due to overcrowding and resource scarcity (Gong et al., 2012; He et al., 2022). Additionally, as population density increases, competition for limited resources intensifies, potentially leading to heightened feelings of social injustice and psychological pressure (Želinsk\u0026yacute; et al., 2021).Building Density: Within the range of 0-0.1, building density is positively correlated with emotional health; within 0.1\u0026ndash;0.2, the impact is minimal; beyond 0.2, the influence becomes negative but weak. Thus, it can be determined that a building density of 0.2 is the critical point, where it contributes most positively to residents' emotional health. As building density continues to increase beyond 0.2, according to urban crowding theory, residents may begin to feel the congestion and stress of urban life, which could lead to a reduction in personal space, invasion of privacy, and competition for public resources (Shan et al., 2022), all of which could negatively impact residents' emotional health, although this negative effect is relatively weak. According to a study by Huang et al. (2024) in Hong Kong, it may be because areas of high density offer better infrastructure quality and convenient living conditions, which may compensate to some extent for the negative impacts of crowding (Huang et al., 2023).Land Use Mix: Above 0.02, land use mix is positively related to emotional health. This may be because increased mix implies greater convenience of services and facilities, thereby enhancing the quality of life (Murgante et al., 2024; Moreno et al., 2021).Road Network Density: Within the range of 0-0.003, road network density has a weakening effect on emotional health; between 0.003\u0026ndash;0.01, emotional health levels improve; beyond 0.01, the impact turns negative. This may be because moderate road network density provides better travel convenience, but excessively high density could lead to negative effects due to traffic congestion and environmental issues.NDVI : Within the interval of 0-0.23, NDVI is positively correlated with emotional health; beyond 0.23, the impact diminishes. This could be because higher NDVI values indicate more green vegetation, offering psychological recovery spaces, but excessively high vegetation cover might bring about negative effects such as dim sunlight, insect bites, and pollen allergies (Maas et al., 2009).\u003c/p\u003e\u003cp\u003eCommercial Service Centrality: Under most circumstances, the impact of commercial service centrality on residents' emotional health is weak. This may be attributed to the widespread adoption of e-commerce and online services in our country over the past decade, which has diminished residents' reliance on physical commercial services. However, when commercial service centrality is low, it has a positive effect on emotional health, likely associated with natural scenic areas. These areas typically exhibit lower commercial service centrality due to certain restrictions on commercial development, yet they may reap positive emotional health benefits because of access to beautiful natural scenery and a peaceful environment (Yuen et al., 2020).Indoor Recreational Centrality: In regions near zero, the absence of indoor recreational facilities is correlated with lower emotional health values; starting from 5, emotional health levels rise with the addition of indoor recreational facilities, stabilizing around 30. The lack of indoor entertainment facilities is associated with lower emotional health values, possibly because the absence of such facilities may lead to a loss of social capital and a dearth of cultural life (Aram et al., 2019), thereby limiting the richness of cultural experiences and spiritual life (Lin \u0026amp; Liu, 2024). Subsequently, as indoor recreational facilities increase, emotional health levels begin to plateau, likely because residents' needs for such facilities are gradually satisfied, and the marginal benefit of new facilities to emotional health diminishes. Moreover, residents may start to pay more attention to the personalization and quality of entertainment facilities, rather than just an increase in quantity.\u003c/p\u003e\u003cp\u003eGreen Space Centrality: In areas close to zero, the lack of green spaces is associated with lower emotional health values; starting from around 1, residents' emotional health increases with the addition of green spaces, stabilizing at around 10. The main findings indicate that in areas with low green space centrality, people lack recreational places and social opportunities. Subsequently, emotional health rise with the increase in green spaces, a phenomenon that can be attributed to the positive effects brought by nature and green areas. For instance, the renowned \"20-minute park effect\" proposed by Yuen et al. (2020) suggests that an increase in green spaces provides residents with more leisure and relaxation spots. Visiting urban parks, whether for a walk or just sitting, as long as there is intimate contact with nature for over 20 minutes, can significantly reduce stress hormone levels. However, after the critical value of 10, the improvement in emotional health begins to plateau, likely because residents' needs for green spaces are gradually met, and the marginal benefit of green spaces to emotional health starts to diminish.Transportation Station Density: Within the range of 0 to 10, there is a positive correlation between transportation station density and emotional health levels; beyond 10, a negative trend emerges. This could be because increased transportation stations enhance travel convenience, but an excessively high density, especially too many private car parking spots, may have negative impacts due to space occupation and environmental issues (Lunke, 2020).\u003c/p\u003e\u003cp\u003e3.2.4 The nonlinear relationship between the interaction effects of different built environment elements and residents' emotional health\u003c/p\u003e\u003cp\u003eFinally, this study delves into the nonlinear impacts of the interactive effects between various built environment elements on residents' emotional health using PDP. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e7\u003c/span\u003e utilizes SHAPley values to demonstrate the interactive effects between pairs of built environment variables. In each plot, the X-axis represents the variable of interest, with the coordinate value indicating the magnitude of that variable. The right Y-axis indicates the variable that interacts most strongly with the variable of interest, with its color representing the magnitude of that variable's value. Each point on the plot represents a sample point, with the color of the point corresponding to the value on the right Y-axis. Concurrently, the positive and negative SHAPley values on the left Y-axis signify the correlation between the feature variables and the target variable; values greater than zero indicate a synergistic effect, meaning that the two variables jointly promote an increase in the dependent variable; whereas values less than zero indicate an antagonistic effect, meaning that when the two feature variables act together, they have a negative impact on the target variable (such as emotional health). The distance on the Y-axis represents the significance of the correlation, with greater distance indicating stronger significance.\u003c/p\u003e\u003cp\u003eThrough these plots, we can more intuitively comprehend how different built environment characteristics interact with each other and subsequently influence residents' emotional health.For instance, the interaction effect plot between \"Building Density\" and \"Population Density\" reveals that as building density increases, the SHAP values show a downward trend, indicating that an increase in population in areas of high building density does not bring benefits to emotional health; instead, it may lead to antagonistic effects due to intensified competition for resources or overcrowding. The interaction effect plot between \"Building Density\" and \"NDVI\" shows that as the \"Buildings\" variable increases, the SHAP values also exhibit a downward trend, suggesting a potential antagonistic effect between the two, which means that an increase in NDVI may not significantly enhance residents' emotional health in areas with high building density. This finding suggests that merely increasing green space may not be sufficient to improve residents' emotional health, especially in areas where building density is already high. In contrast, the interaction effect plot between \"Green Space Centrality\" and \"Commercial Service Centrality\" shows that as \"Commercial Service Centrality\" increases, the SHAP values exhibit an upward trend, implying that improved accessibility to commercial facilities may positively affect residents' emotional health, particularly in environments with a higher degree of outdoor greenery. This synergistic effect may stem from commercial activities providing convenient living services for residents, while outdoor green spaces offer areas for relaxation and leisure, and the combination of the two may create a more livable environment.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis paper explores the complex relationship between the urban built environment and residents' emotional health from the perspective of a 15-minute living circle, using an interpretable machine learning model that integrates XGBoost and SHAP. Among the nine types of built environment features, transportation stations and green spaces were found to have the most significant impact on residents' emotional health. The article argues that transportation stations, such as subway stations, bus stations, and motor vehicle parking stations, as well as green spaces, such as parks, green squares, and natural landscapes, significantly impact residents' emotional health in the built environment because they touch on two fundamental dimensions of urban life: efficiency and resilience (vitality). Transportation stations are the lifeblood of daily urban life. They not only shorten physical distances but also alleviate time pressure, allowing residents to arrange their daily activities more flexibly (Li et al., 2020; Duan et al., 2023), thus demonstrating great importance. However, it is worth noting that overly dense traffic stations, especially the excessive increase in private stations, can lead to parking difficulties, squeezing and occupying space that originally belonged to greenery or transportation, which can have a significant negative effect. Green spaces are urban oases, and their positive effects on residents' emotional health are reflected in their restorative and therapeutic properties. They provide a way to reconnect with nature, help alleviate the stress and fatigue of urban life, stimulate positive emotions, promote relaxation and recovery, which is crucial for residents' emotional health (Woo et al., 2009; Xie et al., 2021).\u003c/p\u003e\u003cp\u003eThe urban built environment exhibits nonlinear relationships and threshold effects with residents' emotional health. This relationship indicates that there is a saturation point to the impact of increasing points of interest in the built environment on residents' emotional health; beyond this point, further increases have limited effects on enhancing emotional health and may even lead to negative outcomes. However, previous studies largely assumed a linear impact of the urban built environment on residents' emotional health, implying that changes in environmental factors would proportionally affect emotional health. The complexity of real-life scenarios suggests that such relationships are not always so straightforward and linear.For instance, when building density is between 0 and 0.1, there is a positive correlation between building density and emotional health values; between 0.1 and 0.2, there is almost no impact on emotional health; and beyond 0.2, building density has a negative effect on emotional health. The findings of this study can help urban planners and policymakers to more accurately understand the mechanisms by which the urban built environment affects residents' emotional health. By identifying and leveraging these nonlinear relationships and threshold effects, they can allocate resources rationally to maximize benefits in enhancing residents' emotional health. The interactive effects between environmental variables reveal which combinations of environmental characteristics may more effectively promote residents' emotional health, suggesting that urban planners need to consider multidimensional environmental characteristics and their interactions when designing urban spaces, to create a healthier and more livable urban environment.By taking these factors into comprehensive account, urban planning can not only enhance the material quality of life for residents but also promote their psychological well-being, thus achieving sustainable urban development.\u003c/p\u003e\u003cp\u003eWith the advancement of urbanization and the improvement of living standards, residents' demands for their living environment and quality of life are increasing. However, traditional urban planning, often dominated by functional zoning, has led to the centralized distribution of various living service facilities in cities. This results in longer travel times and higher costs for residents to access these facilities, thereby reducing their quality of life (Nie, 2024). The 15-minute living circle, where most urban residents conduct their daily activities, can significantly influence psychological feelings, activity willingness, and community atmosphere through the quality of its spatial elements. High-quality, humanized street design can effectively promote positive emotions (Fang, 2015; Murgante et al., 2014). In current urban areas, the increasing pace of life and work pressure, combined with uniform reinforced concrete buildings, not only leads to aesthetic and psychological fatigue but also has adverse effects on people's physical and mental health (Tsui, 2008). This phenomenon has prompted many urban residents to seek comfort and rejuvenation outside the city, with some even taking action to do so. For example, a short resignation letter that went viral in China a few years ago, \"The world is so big, I want to see it,\" inspired many to follow suit. However, seeking restorative experiences or traveling outside the city for extended periods is not always feasible, as people need to work to make a living or stay behind to take care of their families (Lin \u0026amp; Liu, 2024). This raises an important question: How can individuals find opportunities for rest and rejuvenation in their daily lives? The Citywalk, which gained popularity in China in 2024, may provide a possible answer. Citywalk emphasizes exploring urban spaces on foot and immersing oneself in the local customs and development pulse of the city. This reflects a shift from escaping the city to immersing oneself in it. Urban residents no longer see the city solely as a place for work and life but begin to seek deeper connections with it, rediscovering and appreciating its charm through immersive urban experiences. This also indicates that cities are not only places for daily work but also the core space for residents' daily life and entertainment. Therefore, future urban construction and renewal should not only meet daily life and basic work needs but also consider promoting restorative and relaxation experiences, such as adding suitable natural spaces, leisure and entertainment facilities, and rich, diverse street designs, to help residents find a balance and source of vitality in the busy pace of urban life, thereby improving overall quality of life and happiness.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eThis study used an interpretable machine learning model that integrates XGBoost and SHAP, combined with Weibo social media data and geographic big data, to conduct an in-depth analysis of the relationship between the built environment and residents' emotional health in Nanning city. The following conclusions can be drawn:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eFirstly, the overall emotional health level of urban residents in Nanning is relatively good. In terms of temporal distribution characteristics, residents' emotional health values exhibit significant regular changes within a week. Emotional values on rest days are higher than those on weekdays, with the highest values on Sundays, which may be related to increased leisure and relaxation activities during weekends. The lowest emotional point occurs on Thursdays, likely due to fatigue from continuous work. Over the course of a year, emotional health values peak during holidays such as New Year's Day and the Spring Festival, but decrease after holidays like in February and tend to fluctuate steadily in the subsequent months, indicating the positive impact of the holiday atmosphere on emotional health. Regarding spatial distribution characteristics, compared to low and median emotional areas, high emotional areas are more dispersed and have a more pronounced distribution in the peripheral areas of the city, while median and low emotional areas are concentrated and scattered in the central urban area.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eSecondly, regarding the relative importance and characteristic effects of built environment variables, research has found that traffic station density and green space centrality are the two most important factors affecting residents' emotional health. The convenience of transportation stations has significantly enhanced the convenience of daily commuting, reducing time costs and economic burdens. However, it is worth noting that the excessive increase of overly dense transportation stations, especially private stations, can have a significant negative effect. As urban \"oases,\" green spaces provide residents with a way to reconnect with nature, helping to alleviate the pressure and fatigue of urban life and stimulate positive emotions.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThirdly, regarding the nonlinear relationship and threshold effects between the built environment and emotional health, the study finds that factors such as population density, building density, and land use mix are not simply linearly related to emotional health but exhibit threshold effects. At certain thresholds, emotional health levels are optimized, and beyond those thresholds, emotional health levels may decline. For instance, we observe a nonlinear relationship between transportation station density and emotional health. In areas with lower transportation station density (within the range of 0 to 10), there is a positive correlation between transportation station density and emotional health levels. When transportation stations exceed 10, there is a stable negative trend between transportation station density and emotional health levels. The interactive effects indicate which combinations of environmental characteristics may more effectively promote residents' emotional health, and these results also highlight the importance of considering multidimensional environmental features and their interactions in urban planning to achieve an improvement in residents' emotional health and sustainable urban development.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eFinally, there are several limitations in this study. Firstly, the Weibo data used in the study only reflects the emotions of users who use Weibo and like to post on Weibo, which makes the representative sample may be insufficient. Secondly, due to data limitations, the study was unable to control for other social attribute variables that may affect emotional health levels, such as economic status, education level, and age distribution. This resulted in the study only being able to explore the correlation between the built environment and emotional health to a certain extent, without delving into causal relationships. Thirdly, the study did not differentiate the distribution differences of built environment data in different regions of the city, such as the inner and outer urban areas. Due to differences in sensitivity and demand for built environment elements such as traffic density and green space among residents in different areas of the city, for example, suburban areas may be more sensitive to traffic station density, while residents in urban areas may be more concerned about green space. This difference needs to be more fully considered and analyzed in future research.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003eThe data that support the findings of this study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosure of interest\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003eThe authors declare no conflicts of interest in the research presented in this article.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding information:\u003c/strong\u003eThis work was supported by the National Social Science Fund of China (No.22BSH109).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003emeng Cai: Writing\u0026ndash; review \u0026amp; editing, Validation, Supervision;Resources and funds;Xiaoyin Zhang: Writing\u0026ndash;original draft, Visualization, Methodology. Investigation;xue Gong: Writing\u0026ndash; review \u0026amp; editing, Validation, Supervision, Conceptualization.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAllam, Z., Bibri, S. E., Chabaud, D., et al. (2022). The \u0026lsquo;15-Minute City\u0026rsquo; concept can shape a net-zero urban future. \u003cem\u003eHumanities and Social Sciences Communications\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), 126.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAnderson, K., \u0026amp; Smith, S. J. (2001). Emotional geographies. \u003cem\u003eTransactions of the Institute of British geographers\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(1), 7\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAram, F., Solgi, E., \u0026amp; Holden, G. (2019). The role of green spaces in increasing social interactions in neighborhoods with periodic markets. \u003cem\u003eHabitat International\u003c/em\u003e, \u003cem\u003e84\u003c/em\u003e, 24\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBedimo-Rung, A. L., Mowen, A. J., \u0026amp; Cohen, D. A. (2005). The significance of parks to physical activity and public health: a conceptual model. \u003cem\u003eAmerican Journal of Preventive Medicine\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(2-supp-S2), 159\u0026ndash;168.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCervero, R., \u0026amp; Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and design. Transportation research part D: Transport and environment, 2(3), 199\u0026ndash;219.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen Chongxian, Z., Danting, X., \u0026amp; Yu,\u0026amp;Li Shuhua. (2018). Research on the Impact of Urban Street Landscape Features on Human Emotional Health Urban architecture (9), 4.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, T., \u0026amp; Guestrin, C. (2016). XGBoost: a scalable tree boosting system. KDD'16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; 2016 Aug 13\u0026ndash;17; San Francisco, California.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen Zheng, W., Hangbin, \u0026amp; Yu Jue. (2016). \u0026amp; Empirical study on environmental reality perception for urban design Southern Architecture, (4), 10\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChina, E. (2023). Market Data Report\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.100ec.cn/.2023\u003c/span\u003e\u003cspan address=\"https://www.100ec.cn/.2023\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChina Weather Network (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.weather.com.cn/\u003c/span\u003e\u003cspan address=\"https://www.weather.com.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eC40 Cities Climate Leadership Group (2020). How to build back better with a 15-minute city. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.c40knowledgehub.org/s/article/How-to-build-backbetter-with-a-15-minute-city\u003c/span\u003e\u003cspan address=\"https://www.c40knowledgehub.org/s/article/How-to-build-backbetter-with-a-15-minute-city\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed 30 November 2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDuan Zhengyu, H., Xuan, Chen, \u0026amp; Chuan (2023). A Study on the Impact of Urban Residential Location and Travel Behavior on the Quality of Life of Commuters: A Case Study of Shanghai Modern Urban Studies (06), 88\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEwing, R., \u0026amp; Cervero, R. (2010). Travel and the built environment: A meta-analysis. \u003cem\u003eJournal of the American planning association\u003c/em\u003e, \u003cem\u003e76\u003c/em\u003e(3), 265\u0026ndash;294.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFang Rong The spatial characteristics and planning design methods of living streets in 2015 Urban Issues 12:46\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGong, P., Liang, S., Carlton, E. J., Jiang, Q., Wu, J., Wang, L., \u0026amp; Remais, J. V. (2012). Urbanisation and health in China. \u003cem\u003eThe lancet\u003c/em\u003e, \u003cem\u003e379\u003c/em\u003e(9818), 843\u0026ndash;852.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuangxi Nanning Municipal People's Government Portal Website( (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nanning.gov.cn/\u003c/span\u003e\u003cspan address=\"https://www.nanning.gov.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuo Xiaoyan, W. Zhenhong (2007). Concept, Function, and Significance of Positive Emotions. \u003cem\u003eAdvances in Psychological Science\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(5), 810\u0026ndash;815.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuo Yuting and Du Hongwu (2023). Analysis of Spatial Elements of Emotional Health oriented Lifestyle Streets: A Case Study of Guangzhou City Architecture and Culture (6), 122\u0026ndash;124.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHan Huiran, X., \u0026amp; Lingyi,\u0026amp;Yang Chengfeng. (2024). The impact of multi-scale built environment on the mental health of the elderly: an empirical study based on the extreme gradient boosting model in Hefei Geographic Research, 43 (6), 1502\u0026ndash;1521.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHandy, S., Cao, X., \u0026amp; Mokhtarian, P. (2005). Correlation or causality between the built environment and travel behavior? Evidence from Northern California. \u003cem\u003eTransportation Research Part D: Transport and Environment\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(6), 427\u0026ndash;444.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHe, D., Miao, J., Yi, L., Song, Y., Chen, L., \u0026amp; Liu, Y. (2022). Urban greenery mitigates the negative effect of urban density on older adults\u0026rsquo; life satisfaction: evidence from Shanghai, China. \u003cem\u003eCities\u003c/em\u003e, \u003cem\u003e124\u003c/em\u003e, 103607.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHe, P., Yu, B., Ma, J., et al. (2024). Exploring the non-linear relationship and synergistic effect between urban built environment and public sentiment integrating macro-and micro-level perspective: a case study in San Francisco. \u003cem\u003eFrontiers in Psychology\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e, 1276923.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHeilig, G. K. Prospects:the 2022 Revision(United Nations,Department of Eeonomic and Social Affairs,Population Division,2012)[EB/OL]\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e.[2023-08-15].http://esa.un.org/wpp/ppt/paa/PAA_2012_Heilig.pdf\u003c/span\u003e\u003cspan address=\"http://.[2023-08-15].http://esa.un.org/wpp/ppt/paa/PAA_2012_Heilig.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHematian, H., \u0026amp; Ranjbar, E. (2022). Evaluating urban public spaces from mental health point of view: Comparing pedestrian and car-dominated streets. \u003cem\u003eJournal of transport \u0026amp; health\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e, 101532.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuai, S., \u0026amp; Van de Voorde, T. (2022). Which environmental features contribute to positive and negative perceptions of urban parks? A cross-cultural comparison using online reviews and Natural Language Processing methods. \u003cem\u003eLandscape and Urban Planning\u003c/em\u003e, \u003cem\u003e218\u003c/em\u003e, 104307.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang, J., Cui, Y., Li, L., Guo, M., Ho, H. C., Lu, Y., \u0026amp; Webster, C. (2023). Re-examining Jane Jacobs\u0026rsquo; doctrine using new urban data in Hong Kong. \u003cem\u003eEnvironment and Planning B: Urban Analytics and City Science\u003c/em\u003e, \u003cem\u003e50\u003c/em\u003e(1), 76\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHosford, K., Beairsto, J., \u0026amp; Winters, M. (2022). Is the 15-minute city within reach? Evaluating walking and cycling accessibility to grocery stores in Vancouver. \u003cem\u003eTransportation Research Interdisciplinary Perspectives\u003c/em\u003e, 14, Article 100602.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang, Z., Wu, C., \u0026amp; Chung, H. (2025). The 15-minute community life circle for older people: Walkability measurement based on service accessibility and street-level built environment\u0026ndash;A case study of Suzhou. \u003cem\u003eChina Cities\u003c/em\u003e, \u003cem\u003e157\u003c/em\u003e, 105587.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJianxiao, L., Han, B. I., \u0026amp; Wang, M. (2020). Using multi-source data to assess livability in Hong Kong at the community-based level: A combined subjective-objective approach. \u003cem\u003eGeography and sustainability\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e(4), 284\u0026ndash;294.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi Tao, W., \u0026amp; Jiaoe (2020). Happy River Research on the Differences in Inter city Travel Networks between Chinese Residents' Working Days and Holidays [J]. \u003cem\u003eJournal of Geography\u003c/em\u003e, \u003cem\u003e75\u003c/em\u003e(04), 833\u0026ndash;848.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi Xinyang, H., \u0026amp; Zaisheng (2015). Yuan Yiqian, Zhao Lizhi\u0026amp;Xu Minjie Research on the suitability of community public spaces for urban elderly and children Urban Development Research (05), 104\u0026ndash;111.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLin, Y. X., \u0026amp; Liu, Y. (2024).. Let the city heal you: environment and activity's distinct roles in leisure restoration and satisfaction. \u003cem\u003eCities\u003c/em\u003e, 154.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu Zhilin, C. (2001). Yanwei The spatial structure of weekend leisure activities for Shenzhen citizens Economic Geography (04), 504\u0026ndash;508.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., \u0026hellip; Lee,S. I. (2019). Explainable AI for trees: From local explanations to global understanding.arxiv preprint arxiv:1905.04610.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLunke, E. B. (2020). Commuters\u0026rsquo; satisfaction with public transport. \u003cem\u003eJournal of Transport \u0026amp; Health\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e, 100842.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMaas, J., Spreeuwenberg, P., Van Winsum-Westra, et al. (2009). Is green space in the living environment associated with people\u0026rsquo;s feelings of social safety? \u003cem\u003eEnvironment and Planning A\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e(7), 1763\u0026ndash;1777.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMacDonald, G. (2014). Bodies moving and being moved: Mapping affect in Christian Nold's Bio Mapping. \u003cem\u003eSomatechnics\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e(1), 108\u0026ndash;132.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMinistry of Housing and Urban-Rural Development of the People's Republic of China(2018).\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.mohurd.gov.cn/\u003c/span\u003e\u003cspan address=\"https://www.mohurd.gov.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoreno, C., Allam, Z., Chabaud, D., et al. (2021). Introducing the 15-Minute City: Sustainability, resilience and place identity in future post-pandemic cities. \u003cem\u003eSmart Cities\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e(1), 93\u0026ndash;111.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurgante, B., Patimisco, L., \u0026amp; Annunziata, A. (2024).. Developing a 15-minute city: a comparative study of four italian cities-cagliari, perugia, pisa, and trieste. Cities, 146(Mar.), 1.1\u0026ndash;1.22.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurthy, D., Gross, A., \u0026amp; Pensavalle, A. (2016). Urban social media demographics: An exploration of Twitter use in major American cities. \u003cem\u003eJournal of Computer-Mediated Communication\u003c/em\u003e, \u003cem\u003e21\u003c/em\u003e(1), 33\u0026ndash;49.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNanning Municipal Government Website (2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nanning.gov.cn/\u003c/span\u003e\u003cspan address=\"https://www.nanning.gov.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNie Jing (2024). Research on Urban Renewal Strategies Based on the 15 Minute Community Life Circle Concept Residential and Real Estate (16), 91\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eParsa, A. B., Movahedi, A., Taghipour, H., Derrible, S., \u0026amp; Mohammadian, A. K. (2020). \u003cem\u003eToward safer highways, application of XGBoost and SHAP for real-time accident detection and feature analysis\u003c/em\u003e (Vol. 136, p. 105405). Accident Analysis \u0026amp; Prevention.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eP\u0026eacute;rez-Crespo, L., Kusters, M. S., L\u0026oacute;pez-Vicente, M., Lubczyńska, M. J., Foraster, M., White, T., \u0026amp; Guxens, M. (2022). Exposure to traffic-related air pollution and noise during pregnancy and childhood, and functional brain connectivity in preadolescents. \u003cem\u003eEnvironment International\u003c/em\u003e, \u003cem\u003e164\u003c/em\u003e, 107275.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eS\u0026auml;umel, I., Weber, F., \u0026amp; Kowarik, I. (2016). Toward livable and healthy urban streets: Roadside vegetation provides ecosystem services where people live and move. \u003cem\u003eEnvironmental Science \u0026amp; Policy\u003c/em\u003e, \u003cem\u003e62\u003c/em\u003e, 24\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShan Zhuoran, A., Yuehui, Y., \u0026amp; Man (2022). etc Research on the spatiotemporal distribution characteristics and influencing factors of urban residents' emotions based on Weibo big data: taking the main urban area of Wuhan as an example [J]. Urban Development Research, 29 (8): 9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSteptoe, A., Dockray, S., \u0026amp; Wardle, J. (2009). Positive affect and psychobiological processes relevant to health. \u003cem\u003eJournal of personality\u003c/em\u003e, \u003cem\u003e77\u003c/em\u003e(6), 1747\u0026ndash;1776.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVan den Broek, E. L. (2013). Ubiquitous emotion-aware computing. \u003cem\u003ePersonal and Ubiquitous Computing\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(1), 53\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVins, H., Bell, J., Saha, S., \u0026amp; Hess, J. J. (2015). The mental health outcomes of drought: a systematic review and causal process diagram. \u003cem\u003eInternational journal of environmental research and public health\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(10), 13251\u0026ndash;13275.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang Enxu, Z., Jiang, Y., Jun, W., Yicheng, Y., \u0026amp; Pinyan,\u0026amp;Wang Xuejing. (2024). Research on the Impact Mechanism of Built Environment on Urban Vitality Spatial Differentiation at Street Scale Based on MGWR: A Case Study of Shenyang Central Urban Area Geographic Science, 44 (8), 1322\u0026ndash;1331.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWoo, J., Tang, N., Suen, E., et al. (2009). Green space, psychological restoration, and telomere length. \u003cem\u003eThe Lancet\u003c/em\u003e, \u003cem\u003e373\u003c/em\u003e, 299\u0026ndash;300.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXie, Bo, Wang, \u0026amp; Xiao,\u0026amp;Wu Lei. (2021). Research on the Impact of Urban Green Space on Residents' Mental Health Based on Natural Experiments: A Case Study of Wuhan Donghu Greenway Progress in Geographic Sciences, 40 (7), 13.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu, \u0026amp; Dingyuan (2021). \u003cem\u003eLi Xun Parking Problems, Trends, and Countermeasures in Existing Urban Residential Areas [J]\u003c/em\u003e (Vol. 28, pp. 25\u0026ndash;28). Urban Development Research. 06.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu, J. (2013). Research on prediction methods based on online user sentiment analysis. \u003cem\u003eJournal of China Library Science\u003c/em\u003e, (3), 96\u0026ndash;107.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang Jun, L., \u0026amp; Jing (2013). Lang Senyang Subjective assessment of fatigue status and emotional state of foreign enterprise personnel before and after a single workday [J]. \u003cem\u003eChinese Journal of Health Psychology\u003c/em\u003e, \u003cem\u003e21\u003c/em\u003e(05), 683\u0026ndash;685.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYuen, H. K., \u0026amp; Jenkins, G. R. (2020). Factors associated with changes in subjective well-being immediately after urban park visit. \u003cem\u003eInternational journal of environmental health research\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e(2), 134\u0026ndash;145.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eŽelinsk\u0026yacute;, T., Hudec, O., Mojsejov\u0026aacute;, A., \u0026amp; Hricov\u0026aacute;, S. (2021). The effects of population density on subjective well-being: A case-study of Slovakia. \u003cem\u003eSocio-Economic Planning Sciences\u003c/em\u003e, \u003cem\u003e78\u003c/em\u003e, 101061.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, L. (2023). Research on Parking Problems and Countermeasures of Vehicles in Old Urban Residential Areas Urban Architecture (24), 136\u0026ndash;138.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, S. T., Wang, F. F., Duo, F., \u0026amp; Zhang, J. L. (2018). Research on the majority decision algorithm based on WeChat sentiment classification. \u003cem\u003eJournal of Intelligent \u0026amp; Fuzzy Systems\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e(3), 2975\u0026ndash;2984.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Urban built environment, Weibo social media data, Emotional health, Interpretable machine learning, X-minute city life circle","lastPublishedDoi":"10.21203/rs.3.rs-7774299/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7774299/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWith the advancement of urbanization and the improvement of living standards, residents' demands for living environments and quality of life are increasingly high. The built environment not only shapes the daily living space of residents but also significantly impacts their emotional well-being. This study, from a grid-based micro perspective, integrates geospatial big data and social media data from Weibo (Chinese Twitter), employing an interpretable machine learning model that combines XGBoost and SHAP to conduct an in-depth analysis of the complex relationship between the built environment and residents' emotional health in Nanning City. The study finds that, first, residents' emotional health exhibits distinct temporal and spatial distribution characteristics; second, transportation stations and green spaces are the two environmental variables that most significantly affect residents' emotional health; third, there is a nonlinear relationship and threshold effect between built environmental elements and residents' emotional health, indicating that the impact of built environment elements on emotional health tends to stabilize or reverse after reaching certain thresholds; fourth, there are interactive effects among different built environmental elements, suggesting that certain combinations of environmental elements may have a more pronounced impact on residents' emotional health. These findings also highlight the importance of considering multidimensional environmental characteristics and their interactions in urban planning to enhance residents' emotional health and achieve sustainable urban development. The innovation of this study lies in the combination of Weibo big data and geospatial big data, using an interpretable machine learning model to precisely capture the distribution characteristics of residents' emotions at a small-scale spatial level and explore the complex relationship with the 5D built environment. This provides a basis for optimizing the layout of urban built environment elements from the perspective of residents' emotions and holds significant theoretical and practical significance for urban planning and health management.\u003c/p\u003e","manuscriptTitle":"Decoding Urban Emotions: Exploring the Association between Urban Built Environment and Residents' Emotional Health Using Interpretable Machine Learning Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-27 11:39:14","doi":"10.21203/rs.3.rs-7774299/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1376af79-1d1d-48b2-b6e7-56ff2043ef91","owner":[],"postedDate":"October 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-02T10:42:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-27 11:39:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7774299","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7774299","identity":"rs-7774299","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}