https://orcid.org/0000-0001-8627-1239
Figures
Abstract
Under a human-centered approach, accurately identifying the spatial patterns of urban vitality and revealing the mechanisms through which the built environment affects it can scientifically guide the organic cultivation of urban vitality. In light of this, the main urban area of Yantai City is taken as a case study, utilizing multi-source geographic big data to conduct both theoretical and empirical research. An index system for the urban built environment is established based on four dimensions: human perception, functional, accessibility, and building form. Advanced methods, including Deep Fully Convolutional Neural Networks (SegNet), Random Forest Regression (RFR), and Spatial Lag Regression (SLR), are employed to explore the impact of the built environment on urban vitality. The research findings indicate that: (1) Urban vitality presents a composite spatial structure that embodies both “multi-center” and “clustered” characteristics, exhibiting two primary types of local spatial autocorrelation: “high-high” clustering and “low-low” clustering. (2) The disparities in urban vitality reflect an imbalance in the distribution of functional, accessibility, building form, and human perception, with functional playing a more critical role in nighttime and daytime urban vitality than other dimensions. (3) The effects of the built environment on daytime and nighttime urban vitality show varying degrees of heterogeneity regarding significance and direction. Factors such as BPOI(Commercial Points of Interest), integration, accessibility, and vibrancy have a substantial positive impact on vitality clustering, while human perception becomes increasingly important for enhancing nighttime vitality. These results provide refined technical support for urban micro-renewal, enhancing the relevance and effectiveness of response strategies.
Citation: Zhang Y, Wang X, Wang L, Zhang Y, Ye Y, Jiang H, et al. (2026) Measurement of urban vitality and the influence mechanism of the built environment on it based on multi-source data: A case study of Yantai City. PLoS One 21(3): e0343003. https://doi.org/10.1371/journal.pone.0343003
Editor: Ye Wei, Northeast Normal University, CHINA
Received: April 28, 2025; Accepted: January 31, 2026; Published: March 5, 2026
Copyright: © 2026 Zhang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting information files.
Funding: This research was funded by the Youth Innovation Team Project in Universities of Shandong Province (2022RW026); the National Natural Science Foundation of China (42377207, 42207553); the Shandong Provincial Humanities and Social Sciences Project (); the general project of Undergraduate Teaching Reform in Shandong Province (Z2021177); the Shandong Taishan Scholar Young Expert Program(tsqn202306240); the Key project of Research and Development Program in Shandong Province (2022RKY07006); the Foundation of School and Land Integration Development in Yantai (2021XDRHXMQT18); the open foundation of State Key Laboratory of Lake Science and Environment (2022SKL005); the open foundation of State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, CAS (SKLLQG2024); the Humanity and Social Science Foundation of Ministry of Education of China(19YJCZH171); the Youth Innovation Team Project for Talent Introduction and Cultivation in Universities of Shandong Province.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Urban vitality is an important indicator for assessing urban development and competitiveness, as it largely reflects the quality of life for urban residents and the level of high-quality urban development. It has become a hot topic of interest in urban geography and urban planning [1–3]. The United Nations has proposed Sustainable Development Goals (SDG), among which SDG 11 aims to “make cities and human settlements inclusive, safe, resilient, and sustainable”, serving as a crucial objective for enhancing urban vitality. Meanwhile, the focus on the urban built environment has shifted from “morphology-oriented” to “life-oriented”, and the implementation and regulation of urban design have transitioned from a “two-dimensional plane” to a “three-dimensional space”. The approach of “human-centered scale, three-dimensional analysis perspective, and refined control as a means” is an inevitable requirement for the overall built environment in China to enter a stage of humane and refined development. However, the rapid urbanization during the era of incremental development has led to imbalances and disorders within urban development, resulting in phenomena such as inefficient and scarce spatial function layouts coexisting, urban decay in old cities while new districts remain uninhabited, traffic congestion accompanied by the separation of residential and working areas, as well as social stratification and insufficient inclusiveness. These issues have alienated and weakened the orderly distribution of urban vitality. In the context of cities transitioning from incremental outward development to the intrinsic renewal of existing stock, accurately identifying the spatial patterns of urban vitality and deeply revealing the mechanisms through which built environment factors influence it is of great significance for optimizing land layouts, improving infrastructure construction, upgrading built-up area functions, promoting urban spatial renewal, and enhancing the quality of urban spaces.
In 2021, the “14th Five-Year Plan and the Outline of Long-term Goals for 2035” proposed the “implementation of urban renewal actions”. In the same year, the Ministry of Housing and Urban-Rural Development issued a notification regarding the first batch of urban renewal pilot projects, with Yantai being selected as one of the first pilot cities nationwide. In this context, this paper attempts to take the main urban area of Yantai as a case study, measuring urban vitality based on multi-source data such as Baidu heat maps and street view images, focusing on the spatiotemporal attributes of urban residents’ daily activities. Building on this foundation, we expand the representation of “5D” built environment elements to create an urban built environment indicator system that includes material space, perceptual space, place space, and flow space elements. We then employ random forest regression models and spatial lag regression models to explore the mechanisms through which the built environment affects urban vitality, providing scientific reference for the fine optimization of human-centered urban vitality.
Literature review
Analysis and measurement of urban vitality
Urban vitality was initially described as the ability of a place to induce active social and economic activities [4], and it has been extensively and deeply researched in disciplines such as geography, landscape architecture, urban planning, architecture, and sociology [5–9]. The concept was first introduced by American urban planner Jacobs, who believed that urban vitality stems from the diversity of activities and living spaces intertwined by people within the city. Lynch emphasized in urban morphology that urban vitality is the “ degree to which urban form supports important human functions, ecological requirements, and capabilities” [10,11]. Gehl and others, based on urban sociology, defined urban vitality as “the activities of people in urban space” [12]. Montgomery defines urban vitality as continuous human flow, convenient amenities, rich cultural activities, and a sense of dynamism, further proposing that a vibrant place is characterized by people coming and going on the streets and engaging in diverse activities using various facilities over a 24-hour period. Urban vitality can be regarded as a city’s competitiveness and attractiveness, measured by the concentration of human activities across time and space [13,14]. As research advances, a growing number of scholars recognize that urban vitality should be a comprehensive concept. Grounded in urban sociological theory, it is defined as the integrated capacity of a city to sustain and develop itself, studied through multiple dimensions—economic, social, cultural, ecological, and others—thus shifting the understanding of vitality from a solely spatial dimension toward a multidimensional framework. Overall, although different studies may have different focuses, there is a consensus that urban vitality is a diverse range of resident activities influenced by urban spatial form and complex environmental interactions.
Early research on urban vitality was mainly focused on theoretical frameworks based on interviews, surveys, and audits [10,13,15–17], at smaller spatial scales such as communities and neighborhoods. Although these methods could provide detailed information on the aggregation of human activities in space, they had limitations such as small observation scopes, restricted sample sizes, and low spatiotemporal accuracy, making it challenging to comprehensively reflect the characteristics and changing patterns of urban vitality at larger scales [14,18]. In order to more comprehensively identify urban vitality, some scholars have constructed indicator systems from social, economic, cultural, and environmental dimensions to evaluate urban vitality comprehensively [7], or compared vitality differences between cities at a regional scale [19]. With the advancement of multi-source urban data and corresponding machine learning and geographic data analysis technologies, large-scale, high spatiotemporal accuracy geographic data provide the possibility to analyze the spatiotemporal characteristics and changes in urban vitality on a broad scale. Compared to traditional survey methods and related urban data, urban big data has strong penetration, wide coverage, and contains richer spatiotemporal and semantic information, making it particularly suitable for characterizing human-scale urban vitality. Based on this, scholars have used mobile phone signaling data [20], GPS data [21], social media check-in data [22], LBS location data [23], Baidu heat map data [24], restaurant review data [4], street view image data [25], and other sources to measure the activity intensity of spatiotemporal units to study urban vitality at different scales [4,20–25]. Currently, most research refers to static urban vitality as an indicative index, but there is a need to further consider the dynamism of geographic data and improve research reliability and effectiveness.
Research on the impact of the built environment on urban vitality
As a material carrier that supports a range of human activities and their impact, the urban built environment determines the potential and dynamics of urban diversity and will also influence the evolution of urban vitality. Currently, a large amount of research discusses the relationship between the urban built environment and urban vitality, constructing indicators for the urban built environment from qualitative and quantitative perspectives and analyzing their impact on urban vitality [26]. Cervero et al. constructed a “3D” quantification system for the built environment, including density, diversity, and design dimensions [27]. Subsequently, Ewing et al. added distance to transit and destination accessibility to supplement a more comprehensive “5D” quantification system for the built environment, which has become the main basis for quantifying built environments domestically and internationally [28]. Based on this, scholars have enriched the evaluation indicator system of the built environment from aspects such as traffic accessibility, functional, material form, socio-economic factors, land use, and neighborhood attributes, further exploring their impact on urban vitality [19,29–31], research has revealed that these factors are all closely associated with urban vitality, yet their relative importance varies across cities [31,32]. Jiang et al. explored the relationship between vitality and the built environment through urban morphology, functionality, and human-scale factors, finding functionality to be most influential, while some human-scale elements exhibited negative effects. In contrast, Xia et al., in their multi-city study on vitality and environment, noted that land use mix and building density had limited impacts, and the key influencing factors differed from city to city [33]. Jiang et al., in their study on Beijing, indicated that tall, large-area, multi-functional buildings have a significantly positive impact on urban vitality [34]. Jin et al. constructed a relatively comprehensive set of built-environment indicators and identified factors such as floor area ratio, POI density, and POI mix as key enhancers of urban vitality, whereas green space was found to somewhat suppress vitality [35]. However, current research focuses on the physical form of the built environment, the impact of objective perception on urban vitality, while neglecting human emotional perception. There is a lack of subjective evaluation of the built environment by humans and a more comprehensive assessment of the built environment from a human-centered perspective. More specifically, from a human-centered perspective, the built environment includes objective perceptions such as street greening, sky landscapes, building facades, pedestrian spaces [36], as well as subjective perceptions that reflect human psychological feelings towards a place being beautiful, boring, lively, safe, prosperous, etc. [4,37,38]. Sense of security, aesthetics, prosperity, and vibrancy are considered positive subjective perceptions by humans [39], while boredom and other negative subjective perceptions indicate the lack or minimization of these features, significantly enhancing the attractiveness of cities to residents [38,40,41]. The above indicators are directly related to urban vitality.
The combination of multi-source urban data and deep learning algorithms provides technical support for a deeper understanding of the interaction between human behavior and subjective perceptions in cities. Domestic and international scholars have utilized various neural network models, based on street view image data, to quantify subjective perceptions categorized into six classes: safety, vibrancy, dullness, wealth, frustration, and beauty. By collecting perception scores from volunteers on a large number of street view images, they trained models to quantify the subjective perceptions of residents in different cities, thereby providing support for various city-related research [42,43]. In recent years, the combination of street scene images and deep learning algorithms has provided a new method of measuring human perception [44,45]. Zhang et al. used a combination of computer vision to identify heat maps and Convolutional Neural Networks to reveal the visual features that affect human subjective perceptions and consequently influence street vitality, exploring the street elements that shape high street vitality [46]. Ogawa et al. utilized deep learning models to estimate 22 different subjective perceptions in the Setagaya area of Tokyo. Their research confirmed the impact of specific landscape elements such as roads, buildings, and vegetation on perceptions, indicating that urban designers and planners should prioritize certain factors when enhancing subjective perceptions [47]. In addition, with the development of computer vision, the SegNet network model has improved the efficiency and accuracy of image semantic segmentation. Many studies use this method to analyze street scene images to quantify human activities, street quality, and environmental comfort [44,48]. This allows the previously overlooked but important component of cities, which is human subjective perception that also has a significant impact on shaping urban vitality, to be quantified and integrated into the analysis framework of urban vitality and built environment.
Research framework, data, and methods
Research framework
To depict the relationship between the built environment and urban vitality, this paper proposes a research framework as shown in Fig 1. Firstly, based on the Baidu heatmap data of the daily activity patterns of urban residents, the urban vitality in the main urban area of Yantai city is measured during day and night. Secondly, a comprehensive indicator system of urban built environment impact factors is constructed, incorporating multiple spatial elements such as human perception, accessibility, functional, and building form. Finally, a Random Forest regression model and a Spatial Lag regression model are utilized to reveal the complex relationship between urban day-night vitality and the influencing factors of the built environment, including the relative importance of related factors and quantitative effects.
Note: The imagery in this map is sourced from the OpenStreetMap Open-source mapping platform (https://openstreetmap.org). This map bears a resemblance to the original imagery but is not identical.
Study area overview
Yantai is located in the eastern part of the Shandong Peninsula, bordering the Yellow Sea and Bohai Sea. It is one of the first coastal cities opened up in China. Currently, it administers 5 districts, 6 county-level cities, and 5 economic and functional zones, with a land area of 13,900 square kilometers and a population of 7.083 million. To focus on the research subjects and objectives and highlight the spatial differentiation patterns of urban vitality between day and night and different spatial scales, this paper selects the main urban area of Yantai City, which has the highest population density, as the study area (Fig 2). For precise identification of human-oriented urban vitality, considering the walkability of the street network and residential areas, 300m × 300m grids are selected as the basic spatial analysis units. According to the administrative division of Yantai in 2023, the study area includes 12 street space units and 3372 grid space units, in order to ensure data completeness for subsequent analysis, we excluded grid cells that had no roads and did not contain any data. After the exclusion, there are a total of 2,365 grid space units remaining.
Note: The imagery in this map is sourced from the OpenStreetMap Open-source mapping platform (https://openstreetmap.org). This map bears a resemblance to the original imagery but is not identical.
Data source and processing
- (1). Basic Urban Spatial Data
Based on Baidu Maps API interface, 37,851 pieces of POI data were obtained, covering 11 categories including accommodation, shopping, dining, residence, transportation, living facilities, sports and leisure, and public facilities. City building vector data and road data are from the OpenStreetMap open mapping platform (https://www.openstreetmap.org/).
- (2). Urban Vitality Data
To measure urban vitality, population heat data, urban POI data, and online review data were selected [49], sequentially measuring “human flow agglomeration-dispersion vitality, activity facility vitality, and network perception vitality.” Finally, principal component analysis (PCA) [50] was employed to comprehensively calculate these three vitality indicators, yielding the ultimate urban vitality as shown in Fig 3. Baidu Heatmap data was collected to characterize population agglomeration levels. From an external manifestation perspective, human flow density is the most direct indicator for measuring vitality, while the built environment of streets constitutes the decisive factor generating human activities—among which functional characteristics form the foundational condition for sustained human flow. From an intrinsic attraction perspective, the network effect of physical spaces embodies the essence of street vitality, and online reviews can reflect subjectively perceived vitality in a bottom-up manner. By using the location information of users of Baidu-related mobile applications (such as Baidu Maps, Baidu Search, Baidu Weather, etc.) and projecting it onto space, the Baidu Heat Index is calculated. Different colors are used to reflect the spatial distribution of users in a certain area, which has been widely used to measure population aggregation characteristics [51]. By accessing the Baidu dynamic API through a web crawler, heat maps of the main urban area of Yantai City for a period of 5 working days from June 13th to 17th, 2022, between 7:00 and 24:00 were collected at 2-hour intervals. The period from 7:00–18:00 was defined as the daytime period, while 18:00–24:00 was defined as the nighttime period, and the urban vitality was separately calculated for daytime and nighttime. Heatmap data is considered point data in ArcGIS, which includes fields such as identification number, longitude and latitude, time, and heat value. This data is imported into ArcGIS, where the Kernel Density Analysis tool is used to analyze the density values of each point within a certain range, resulting in raster data for each time period. Then, the raster data is converted into point data and assigned to the corresponding grid units to create a vitality map for each time period. Finally, the heat values for both daytime and nighttime periods are summed and averaged to obtain the final urban vitality data for day and night. Due to the climate reasons that Yantai City is in early summer in June, making it suitable for outdoor activities by residents, their behavioral activities tend to be average. The data collection took place on sunny days, which eliminates the impact of extreme weather conditions on residents’ mobility, making it suitable for studying through population heat maps.
Note: This illustration is drawn by the author.
- (3). Subjective and Objective Perception Data
Street view images were obtained using Python language by accessing the Baidu Street View Image API with a sampling interval of 50m, generating 1599 sampling points. Street view images were captured from the front, back, left, and right of each sampling point with a 90-degree field of view and 0-degree horizontal line, resulting in a total of 3873 valid street view images. These images were then stitched together into panoramic images using OpenCV programming to provide a comprehensive coverage of street spatial environmental features at a human scale. The SegNet semantic segmentation model was used to identify elements such as sky, sidewalk, lane, building, and greenery within the street space features. Based on this, the proportion of each element in each street view image was calculated to create objective perception variables for human perception (Fig 4). This model’s code is open source(GitHub - CSAILVision/semantic-segmentation-pytorch: Pytorch implementation for Semantic Segmentation/Scene Parsing on MIT ADE20K dataset) and can be directly accessed and used in practice. The author of the code trained the model using the MIT ADE20K dataset, achieving an accuracy of 80.13%.
Note: The imagery in this map is sourced from the OpenStreetMap Open-source mapping platform (https://openstreetmap.org). This map bears a resemblance to the original but is not identical.
The VGG16 deep learning model was utilized to effectively identify and learn features from each street-view image. For this study, 48 volunteers were recruited. To ensure fairness in scoring, an average of three volunteers was selected from each subdistrict in Zhifu District to participate in image evaluation. Among them, 28 were male and 20 were female. In terms of age distribution, 14 volunteers were aged 20–30, 22 were aged 30–40, and 12 were over 40 years old. All volunteers had resided in Yantai for more than eight years and possessed a relatively comprehensive understanding of the city’s basic conditions. In this study, volunteers were gathered to rate 300 street view images based on five aspects: beauty, boring, lively, safe, and wealthy. The ratings were categorized into five levels: high-high, high, medium, low, and low-low. The scores from high to low represent the most subjective feelings of people towards the street environment. The 300 images and their corresponding scores were input into the VGG16 model for recognition training, the accuracy of the trained model is shown in the Table 1. The trained model was then used to rate the remaining 600 street view images, resulting in scores for all street view images. This process forms subjective perception variables for human perception. In Fig 4. we selected a street scene image’s score prediction as an example. It can be seen that the corresponding score for this image is the highest, placing it in the high-score category.
Selection and characterization of built environment influencing factors
By integrating existing research results and incorporating the “5D” model elements of the built environment, the study compares the roles of environmental elements in place space and flow space, supplements with non-material environmental elements perceived by humans, and analyzes the potential existence of specific attribute isolation and differentiation phenomena. This is to more diversely reveal the mechanisms of how the built environment influences urban vitality as a whole, supporting targeted vitality optimization strategies. Among them, the architectural morphology elements are the underlying supportive factors that affect the comprehensive vitality of the city. Buildings provide the foundation for urban construction and development, while building density and floor area ratio offer the capacity for the built environment. Functional elements are direct driving factors influencing urban vitality. Indicators related to urban functions, such as Point of Interest (POI) data, were used, selecting POI density and POI diversity to measure the level of functionality. POI density reflects the quantity of urban infrastructure related to retail, dining, entertainment, education, and healthcare, while POI diversity indicates the level of completeness of regional infrastructure. Accessibility serves as both an inducing and moderating factor affecting urban vitality, with indicators like selection degree, road density, and proximity chosen for representation, which intuitively reflects the smoothness of traffic and the strength of spatial connections. Human perception elements represent subjective human experiences, which can better promote urban vitality. Based on previous research, these were divided into subjective and objective perceptions. Specifically, a 24-factor indicator system was constructed based on four dimensions: human perception, accessibility, functional and building form. The calculation formulas for accessibility, functional, and other indices can be found in Table 2.
Research methods
Kernel density estimation method.
Kernel density estimation is a method that estimates the density distribution of data across the entire spatial range by smoothing the kernel function around each data point [52]. This technique can accurately and intuitively express the clustering characteristics of elements in space, infer the distribution of event density, and demonstrate the degree of spatial aggregation of urban vitality. In this study, kernel density analysis is used to simulate Baidu heat map data continuously, in order to reflect the spatial clustering of users in the region and the spatiotemporal dynamic characteristics of urban vitality. The mathematical expression for kernel density estimation is as follows:
In the formula: n represents the number of data points; h is the bandwidth; (x-xi) represents the distance from the estimated point x to the sample point xi.
Bivariate spatial autocorrelation analysis.
Spatial correlation can be used to assess the relationships between specific areas, certain geographic phenomena, or attribute values and corresponding indicators in neighboring regions [53], thereby reflecting the spatial distribution of geographic phenomena or specific attribute values. To further explore the spatial correlations among multiple variables, Anselin and others proposed the use of bivariate spatial autocorrelation methods to study the spatial relationships of the same variable at different time points [54,55]. In this study, the GeoDa software is employed to conduct bivariate local spatial autocorrelation and reveal the spatial distribution characteristics of urban vitality during day and night. The calculation formula is as follows:
In the formula provided: represents the value of attribute y in spatial unit a.
represents the value of attribute z in spatial unit c.
and
are the average values of attributes y and z.
and
are the variances of attributes y and z, respectively. n is the number of spatial units,
is the spatial weight, defined using the Rook adjacency criterion. Attribute y refers to the exposure index or sensitivity index, z refers to the coping capacity index.
Random Forest Regression model (RFR).
RFR is an emerging and highly flexible machine learning algorithm that has been widely used to solve classification and regression problems [56,57]. Based on decision tree theory, random forest randomly generates multiple decision trees and selects the tree with the highest level of repetition as the final result [58], thereby obtaining the importance ranking of variables [59]. In this study, we builds a random forest model based on the “randomForest” package in R to conduct a regression analysis, examining the importance of various built environment variables on urban vitality. First, install the “randomForest” package in R Studio, import the processed data, and divide it into training and testing sets. Use the training set to construct the random forest model, and then evaluate the model’s accuracy using assessment metrics. Here, RMSE, MAE, R², and % Var explained are selected to assess the model’s precision. Finally, the importance ranking of the variables can be visualized and the results output.
Variation Partitioning Analysis (VPA).
Variation Partitioning Analysis (VPA) is an extension of Redundancy Analysis (RDA) that is used to decompose the total variance or total variability in a dataset to determine the contributions of different factors to the variability of the data. This method helps to reveal the relationships between dependent variables and explanatory variables in the environment, as well as the extent to which different environmental variables influence the dependent variable [60]. In this study, VPA is conducted using the “vegan” package in R language to determine the contribution rates of various elements in the built environment to changes in urban vitality.
Spatial Lag Regression model (SLR).
To address the spatial differences in urban vitality and the complex spatial variations, this study utilizes a linear regression method with spatial autoregressive disturbances known as Spatial Lag Regression Model (SLR) [61] to quantify the spatial autocorrelation of the dependent variable. The formula for the SLR model is as follows:
Where ln(yi) and ln(yj) are the logarithmic values of urban vitality for grid i and grid j respectively, xi is the vector form of the independent variables, β is the vector form of estimated coefficients, wijln(yj) represents the spatial lag term, wij is used to reflect spatial neighborhood relationships, known as the spatial weight matrix; ρ is the estimated coefficient for the spatial lag term, and εi is the random error term.
Analysis results
Analysis of the spatiotemporal pattern of urban vitality
ArcGIS tools were used to perform kernel density analysis on the obtained Baidu heatmap data, and the visualization of urban vitality for daytime and nighttime periods revealed that, overall, the urban vitality in the main urban area of Yantai City showed spatial and temporal dynamic changes of clustering-dispersion-clustering from 7:00–23:00. The distribution of high and low values of vitality during daytime and nighttime periods was spatially similar, indicating a relatively stable day-night variation in vitality (Fig 3).
From a spatial perspective, there are differences in urban vitality in the main urban area of Yantai City along the north-south and east-west axes, with higher vitality in the north and lower in the south, higher in the east and lower in the west, and a high center surrounded by lower surrounding areas. The areas with highly concentrated urban vitality mainly include the northeastern regions such as Tongshen Street, Yuhuangding Street, and Xiangyang Street, as well as along the Dahaiyang Road and Airport Road, and the eastern part of Zhichu Street and the central region of Fenghuangtai Street. This is primarily due to a higher number of residential areas, relatively complete infrastructure, and a relatively full range of commercial and entertainment facilities, which strongly attract urban residents, resulting in a relatively dense population distribution and thus higher levels of urban vitality. At the same time, the urban vitality of the Yantai main urban area is also influenced by topographical factors. The central region, characterized by low vitality values, is significantly affected by the terrain, as it is located in a mountainous area with Huangjinding, Zhenshan, and Jinkuangding. East of Airport Road is Tashan, and the western edge is traversed by the Jia River. This area has a low population density, and daytime vitality values are low due to activities such as hiking and leisure. Nighttime vitality is extremely low. Overall, urban vitality in the main urban area of Yantai City tends to be concentrated in areas with high-density populations, leisure and entertainment hubs, and regions with developed transportation.
From a temporal perspective, the distribution of urban vitality in the main urban area of Yantai is generally similar between day and night, with slight differences in the levels of vitality. The daytime vitality values are relatively higher and have a wider distribution compared to nighttime values. At night, the vitality values show a concentrated pattern, with high-vitality areas almost entirely located in residential neighborhoods. This indicates that the urban vitality of the main urban area of Yantai is significantly influenced by the living and working habits of its residents. Further, a bivariate Moran’s I method is employed to explore the relationship between urban vitality during the day and night. Daytime vitality is chosen as the first variable, and nighttime vitality as the second variable. The results of the bivariate Moran’s I show the spatial autocorrelation of urban vitality at different times. Specifically, the global Moran’s I value for urban vitality between day and night is 0.722, and through significance testing, it indicates that urban vitality during the day and night exhibits significant spatial aggregation characteristics. As shown in Fig 5, urban vitality exhibits a notable high-high clustering phenomenon, indicating that the spatial distribution of urban vitality during the day and night generally follows the same trend. The high-high regions are mainly distributed in areas such as Xingfu, Tongshen, Baishi, Xiangyang, Yuhuangding, and Shihuiyao streets, which are zones with concentrated populations in the main urban area of Yantai. The low-low vitality aggregation areas are primarily located in the western riverbank area along the Jia River, the central mountainous river area, and the northern coastal dock area. Additionally, some grid units exhibiting high-low clustering are primarily concentrated in industrial land areas, aligning with the pattern of residents working during the day and resting at night.
Note: The imagery in this map is sourced from the OpenStreetMap Open-source mapping platform (https://openstreetmap.org). This map bears a resemblance to the original but is not identical.
Analysis of the relative importance of the elements of the built environment
Furthermore, to enhance the rigor of the indicator system and reduce interference with subsequent research, this study employed an ordinary least squares (OLS) model, utilizing tolerance and variance inflation factor (VIF) values to screen variables influencing urban vitality. Variables with tolerance > 0.1 and VIF < 7.5 were selected. Among the 24 initial variables, Choice and Integration were excluded, resulting in 22 retained variables (Table 3).
The random forest algorithm is used to further explore the relative importance of each element in the built environment. The relative importance of independent variables reflects the contribution of each independent variable to the prediction of results during the modeling process, and the sum of the relative importance of all independent variables is 100%. By ranking the indicators in descending order of importance, a higher importance level indicates that the corresponding element in the built environment has a greater impact on urban vitality. Table 4 displays the performance metrics of the random forest regression results based on the validation dataset. Specifically, the RMSE (Root Mean Square Error) and MAE (Mean Absolute Error) are both less than 0.1, indicating that the difference between the predicted results of the random forest model and the actual results is minimal, reflecting a high level of accuracy for the model. Additionally, the R² and % Var explained (overall explanatory power) are both greater than 67%, suggesting that the random forest model has a good fitting degree and can effectively explain the correlation between variables.
The relative importance of various elements in the built environment is measured by the impact of variables on the node mean square error during the node splitting process using the Random Forest algorithm. A bar graph is then plotted to illustrate this importance. From Fig 6, it can be observed that functional factors, building form factors, and accessibility factors all have a certain influence on the spatial distribution of urban vitality levels in the study area, as depicted in the graph.
To validate this result, further variability decomposition analysis was conducted to explore the relative contributions of explanatory variables in the dimensions of functionality, accessibility, building form, and human perception to the spatial distribution of urban vitality. The variability decomposition analysis results indicate that the dimensions of functional factors, building form factors, and accessibility factors together account for the highest percentage of variance, explaining 18.0% of the daytime and nighttime urban vitality spatial distribution (Fig 7).
Overall, functional factors contribute more significantly to the spatial distribution of urban vitality than morphological, accessibility, and perceptual factors. The individual explanatory power of perceptual factors is relatively low, which aligns with the findings of Ye et al.‘s study of Shenzhen, indicating that functional elements are the primary drivers of urban vitality [4]. As time progresses, the importance of factors influencing urban vitality perception categories continues to increase, indicating the need for a more in-depth evaluation of places at a human scale [62].
In vertical spatial distribution, there are differences in the extent to which each factor of the daytime and nighttime built environment affects urban vitality, but BPOI, floor area ratio, building density, and proximity all show relatively high importance, and their effects are statistically significant. Further comparison of the importance of various factors in the daytime and nighttime built environment reveals the following results: (1) Functionality. Among all functional variables, BPOI plays a significantly more crucial role in shaping urban vitality than PPOI, TPOI, IPOI, RPOI, and GPOI. Its relative importance reaches 0.347 during the day and 0.273 at night. The diversity of POIs contributes more to shaping nighttime urban vitality than daytime vitality, which aligns with residents’ daily activity patterns (working during the day and engaging in dining, shopping, and entertainment at night). Liu et al.’s study of Nanshan District, Shenzhen, yielded a similar finding: specifically, that recreational functions significantly boost urban vitality [63]. This suggests that while commercial facilities can enhance daytime vitality to a greater extent, the spatial agglomeration of diverse amenities serves as a crucial measure for promoting nighttime vitality. (2) Building Form. Floor area ratio and building density are important indicators affecting urban vitality, with relative importance of 0.132 and 0.179 during the daytime and nighttime respectively for floor area ratio, and contributions of 0.095 and 0.099 for building density to daytime and nighttime vitality. This indicates that urban vitality is essentially the interaction between residents and the elements of the built environment, representing the ability of a place to provide facilities and spatial elements to meet the residents’ activities, especially in large cities where land for construction is scarce, vertical growth of construction space is more important for urban vitality than horizontal growth. It is worth noting that the relative importance of building density is higher at night than during the day, which is related to the fact that the main urban area is an old town, and many young workers reside in city villages with high building density. (3) Accessibility. Closeness, Betweenness, and road density are three indicators that impact urban vitality. Closeness, in particular, exhibits a relative importance level of 0.074 during the daytime and 0.075 at nighttime. Overall, these five indicators contribute more to nighttime urban vitality than during the day. (4) Human Perception. Subjective variables related to human perception exhibit stronger explanatory power in the evaluation of urban vitality compared to objective variables. Additionally, subjective variables have a greater influence on nighttime vitality, primarily reflected in the variables of liveliness and affluence.
Analysis of the relationship between built environment and urban vitality
The relationship between various factors of the built environment and urban vitality was explored through the global model and spatial lag regression (SLR) model under ordinary least squares (OLS) method. The Moran’s I values for daytime and nighttime urban vitality were 0.735 and 0.745 respectively, showing significant statistical significance at the 0.01 level. Therefore, the spatial lag regression model is suitable for further exploring the mechanisms influencing urban vitality. Diagnostic information of the spatial lag regression model is shown in Table 5, the R² values for both the daytime and nighttime spatial lag regression models are around 0.8, indicating that these models fit well with the relationship between urban vitality and the built environment during both day and night. The RSE (Residual Standard Error) values are 0.103 and 0.094, respectively, which are relatively low, suggesting high model accuracy. Furthermore, the AIC (Akaike Information Criterion) values for both the daytime and nighttime models are also quite low, indicating that these models are suitable for explaining the relationship between independent and dependent variables. Furthermore, the Moran’s I test of SLR residuals showed a random distribution within the 95% confidence interval, indicating that the spatial lag regression model could effectively explain the relationship between urban vitality and its variables.
The statistical results of the coefficients in the spatial lag regression model (Table 6) indicate that the direction of change in the coefficients of various factors of the built environment and their relationship with urban vitality during the day and night show considerable similarity. The safety factor has a relatively small effect in the daytime vitality model, while the GPOI (Green Points of Interest) and greenery ratio factors have a limited impact in the nighttime vitality model. Moreover, these variables also exhibited relatively low importance in the random forest regression analysis.
In terms of accessibility impact, factors such as betweenness, road density, and closeness all have relatively obvious positive effects on enhancing urban vitality. Consistent with previous research conclusions, accessibility is an important indicator for evaluating urban vitality. The convenience of travel strengthens bidirectional communication between the local area and its surroundings, thereby attracting and consolidating urban vitality in that area.
In terms of functional impact, it can be observed that, (1) The most significant difference is that GPOI does not have a noticeable impact on nighttime urban vitality, while it is negatively correlated with daytime urban vitality. This supports the recent research results of Jiang et al. (2024) on quantifying urban form and urban vitality, as well as the conclusion that green patches are negatively correlated with urban vitality. This finding reveals the complex dynamics of green spaces in urban planning. Although green spaces are generally considered to have a positive impact on urban vitality [2], an excessively high proportion of large green parks may have a negative effect on urban vitality. This could be due to the fragmentation and regionalization of urban space caused by a large number of green patches, reducing opportunities for social and cultural interactions among city residents, thus weakening the overall urban vitality. IPOI is positively correlated with daytime urban vitality and negatively correlated with nighttime urban vitality, while RPOI and GPOI exhibit exactly the opposite relationship. This aligns with residents’ daily work habits, where factories or industrial parks operate during the day and rest at night, and residents leave their homes during the day and return home at night, often choosing to go out for leisure and relaxation. (2) BPOI exerts a significant positive effect on urban vitality, while PPOI also demonstrates a certain positive effect, with both exhibiting weaker impacts at night than during the daytime. POI diversity shows a positive influence on urban vitality, aligning with the finding by Sun et al. in their Nanjing study that “the diversity of service facilities has a substantial positive effect on urban vitality [64].” Furthermore, this study reveals that its positive effect is stronger at night than during the day. This is consistent with the importance results from the random forest model, indicating that while BPOI can enhance urban vitality, only mixed-use neighborhoods can maximize urban attractiveness by attracting population, promoting the development of commerce and industry, and providing opportunities for communication and interaction. In general, functional indicators such as BPOI, POI diversity, and PPOI can effectively enhance urban vitality, and these indicators are also of high importance in the random forest model results.
In terms of building form impact, floor area ratio, and building density are strongly positively correlated with urban vitality. Higher floor area ratios and building coverage ratios contribute to reducing vacancy rates, providing a wider range of spatial use options, and increasing the variety and capacity of continuous activities, thereby promoting an increase in urban vitality [4]. In their study of Lanzhou, Zhang et al. also noted that greater development intensity and higher building density are more conducive to enhancing urban vitality [65].
Subjective perceptual variables have both positive and negative correlations with urban vitality, but they are increasingly important for nighttime urban vitality. Boring has a weak negative correlation with urban vitality, indicating that residents tend to go to lively places. Safety has a positive correlation with nighttime street vitality, as a safe nighttime environment can promote nighttime street vitality, making it more impactful at night. Attributes such as liveliness, beauty, and prosperity have a positive impact on both daytime and nighttime urban vitality, suggesting that improving the appearance and prosperity of the streets can be an effective way to enhance urban vitality.
Discussion
Although existing literature has explored the impact of the built environment on urban vitality, this study offers new insights. In urban renewal, to enhance urban vitality, it is essential to identify and understand the spatial and temporal needs of residents’ daily activities, while also ensuring equitable policies based on individuals, time, and location, so as to improve the relevance and effectiveness of urban renewal (Fig 8).
Firstly, compared to other built environment characteristics, the functional dimension is the most crucial aspect in enhancing urban vitality during both day and night. BPOI, POI diversity, and PPOI are the variables contributing the most to various urban functions and are also the focus of urban designers. Wu et al. have shown that in communities with a high proportion of residential land use, diversity in land use functions is essential. Land use diversity refers to the degree of combination of different types of land use in a specific area, and urban vitality is closely related to land use diversity. Mixed land use patterns can create various activity opportunities for residents, businesses, etc., leading to a vibrant urban environment. The positive correlation between land use diversity and urban vitality has been widely recognized in previous studies. For instance, research indicates that communities with high functional diversity are conducive to walking and biking, stimulating more social interaction and stronger community awareness, all of which contribute to enhancing urban vitality [18,28]. From this, it can be seen that, a higher degree of mixed land use can also lead to more effective land use, improve environmental governance, and help establish a more sustainable and livable city [66,67], in light of this, we should actively explore ways to enhance and increase the quality of living facilities in urban planning and construction. For instance, we can carry out comprehensive renovations of old communities such as Happy New Town and urban villages, and improve public service facilities around residential areas, including education, healthcare, and elderly care. Many previous studies have primarily focused on functional diversity as a key indicator of functional metrics, while neglecting to identify which specific functions play the most critical roles among various urban functionalities. Therefore, this study incorporates the density of various functional Points of Interest (POIs) as a metric. The results indicate that in the main urban area of Yantai, Business POIs (BPOIs) are the most significant type of function contributing to urban vitality. This suggests that areas rich in commercial facilities, particularly various commercial centers, can greatly satisfy people’s daily shopping and consumption needs, while also addressing their requirements for dining, leisure, and social interactions. Consequently, these areas attract substantial foot traffic and make the greatest contribution to urban vitality. This implies that in mixed-use urban functions, commercial functions play a dominant role. By developing commercial facilities to attract foot traffic, it can subsequently promote the construction of other functional facilities and drive urban development, thereby maximizing urban vitality. Compared to form characteristics that have long-term impacts on human perception and behavior [68,69], functionality reflects current human needs in the short term. Unlike the physical form of a city, functional features are easier to change in urban design practice. Government agencies and planners can foster new urban functions through urban renewal, allowing for mixed-use development and compatibility of functions, especially in the construction of commercial facilities, generates synergistic effects among these functional areas, thereby meeting people’s diverse needs for physical space.
Spatial accessibility measures the potential of a space to attract transportation access, with higher accessibility leading to stronger urban vitality. It can influence the development patterns, functional layout, and resource allocation of a city by affecting people’s choices of various activity locations, thereby impacting the economic, commercial, cultural, public spaces, and amenities development of the city. In the accessibility indicators, the importance of proximity and choice during the day and night is relatively higher than that of integration and permeability, which is contrary to the findings of Wu et al. This discrepancy may be attributed to the road system in Yantai’s main urban area. Although the road network in Yantai’s main urban area is dense and offers good accessibility, most of the roads have low classifications and there are many T-junctions, resulting in weak connectivity and insufficient ability to attract pedestrian traffic. Therefore, urban planners and decision-makers should prioritize investments in transportation infrastructure and services during urban renewal efforts, further optimizing the road network structure in Yantai’s central urban area. This includes addressing dead-end roads and T-junctions, implementing intersection channelization, enhancing overall traffic capacity, and promoting compact, multifunctional, and walkable communities to improve accessibility and foster urban vitality [33].
Higher building density and floor area ratio promotes economic activities and social interactions in cities. The wider the spatial extent of human activities, the greater the capacity to accommodate more people and buildings, leading to increased urban vitality, which is consistent with previous research findings [3,33]. Therefore, as a pilot city for urban renewal, Yantai should upgrade and renovate the old buildings in traditional commercial districts, adjust their functional usage, encourage the development of high-rise buildings with diverse functions, create varied and vibrant public spaces, and make full use of both the horizontal and vertical space of existing land to enhance urban vitality.
The method of measuring residents’ perceptions from a human-scale perspective provided in this article can offer more accurate implementation suggestions for urban renewal efforts. The street environment related to the subjective and objective variables of the human perception dimension are potential focus areas for enhancing future urban vitality. Integrating street view images to comprehensively evaluate the built environment can help address the need for micro-updates in cities and provide precise support for street environment improvements. Green environments support better experiences and aesthetics for pedestrians; however, urban greening should take residents’ visual experiences into account. Overly abundant greenery can instead lead to a feeling of dullness and insecurity, causing residents to stay away, the site should provide a variety of interface elements while ensuring residents have access to green landscapes, novel visual experiences, and diverse functional experiences to attract pedestrian traffic. Establishing complete and continuous sidewalks and improving local pedestrian-related facility attributes can significantly promote walkability, enhance urban vitality, and increase safety. At night, residents rely more on their perception of the surrounding environment and accessibility, while street safety facilities such as fences and street lights, as well as life service facilities, have a crucial impact on pedestrians’ psychological safety, residents tend to prefer places that are safer, more open, and less enclosed [70]. Therefore, urban planners should focus more on the micro level during urban renewal efforts, further integrating the “people-centered” concept into the process. Enhancing the safety and accessibility of green public spaces is crucial for promoting nighttime vitality.
Conclusion
During the transitional phase from incremental outward expansion to stock-based intrinsic renewal in urban development, precise identification of the spatial patterns of urban vitality under a human-oriented perspective and an in-depth exploration of the underlying mechanisms through which the built environment influences it serve as a foundational prerequisite for scientifically enhancing urban vitality. In light of this, this research, leveraging geospatial big data and related analytical techniques, measured urban vitality based on the spatiotemporal attributes of urban residents’ daily activities. It constructed a comprehensive impact indicator system for the urban built environment, encompassing multiple spatial dimensions, and integrated relevant theoretical and empirical studies. The primary findings are as follows:
- (1). The integrated use of multi-source geospatial big data can identify the spatiotemporal differences between daytime and nighttime urban vitality. Combining the Random Forest model and Spatial Lag Model enables a precise analysis of the complex impact mechanisms of multidimensional built environment elements on urban vitality.
- (2). Urban vitality in Yantai’s central city exhibits a composite spatial structure combining “polycentric” and “cluster-based” patterns, demonstrating significant positive spatial autocorrelation. This is primarily manifested as spatially segregated clusters of “high-high” and “low-low” agglomerations.
- (3). Urban vitality results from the unbalanced allocation of multidimensional elements, including functionality, accessibility, building morphology, and human perception. Functionality is the most critical dimension influencing both daytime and nighttime vitality, with the direction and intensity of influence from different functional factors showing marked heterogeneity between day and night due to variations in activity demands. Within accessibility, road density, choice, and integration positively promote urban vitality. In building morphology, floor area ratio and building density have a significant positive impact on vitality agglomeration, particularly more pronounced at night. Regarding human perception, subjective perception exerts a greater influence on vitality than objective perception, with its role in enhancing nighttime vitality becoming increasingly important.
This article incorporates subjective and objective perception variables of human perception into the study of the impact mechanism of urban vitality. It uses random forest and spatial lag regression models to explore the complex dynamic relationships between various variables of the built environment and urban vitality. This study has significant reference value for expanding the theoretical understanding of urban vitality, improving quantitative analysis methods, and guiding practical applications. In the future, leveraging multidimensional geographic big data to construct a vitality model covering spatiotemporal attributes of activities will help identify the spatial pattern of urban vitality comprehensively. This will open up innovative paths for integrating top-down multiscale studies on urban vitality and provide scientific references for the fine optimization of people-oriented urban vitality.
Supporting information
S1 File. This document contains the data values for all variables in this paper.
The first and second columns represent the data identifiers, the third and fourth columns denote the two dependent variables, and the remaining columns correspond to the independent variables.
https://doi.org/10.1371/journal.pone.0343003.s001
(ZIP)
Acknowledgments
Our deepest appreciation to the anonymous reviewers and editors for their meticulous work and thoughtful suggestions, which have greatly enhanced this paper.
References
- 1.
Jacobs J. The death and life of great American cities; 1961. 458 p.
- 2. Lopes MN, Camanho AS. Public green space use and consequences on urban vitality: an assessment of European cities. Soc Indic Res. 2012;113(3):751–67.
- 3. Meng Y, Xing H. Exploring the relationship between landscape characteristics and urban vibrancy: a case study using morphology and review data. Cities. 2019;95:102389.
- 4. Ye Y, Li D, Liu X. How block density and typology affect urban vitality: an exploratory analysis in Shenzhen, China. Urban Geogr. 2017;39(4):631–52.
- 5. Zhang Z, Liu J, Wang C, Zhao Y, Zhao X, Li P, et al. A spatial projection pursuit model for identifying comprehensive urban vitality on blocks using multisource geospatial data. Sustain Cities Soc. 2024;100:104998.
- 6. Li Z, Zhao G. Revealing the spatio-temporal heterogeneity of the association between the built environment and urban vitality in Shenzhen. IJGI. 2023;12(10):433.
- 7. Yang H, He Q, Cui L, Mohamed Taha AM. Exploring the spatial relationship between urban vitality and urban carbon emissions. Remote Sens. 2023;15(8):2173.
- 8. Lee S, Kang JE. Impact of particulate matter and urban spatial characteristics on urban vitality using spatiotemporal big data. Cities. 2022;131:104030.
- 9. Yue W, Chen Y, Zhang Q, Liu Y. Spatial explicit assessment of urban vitality using multi-source data: a case of Shanghai, China. Sustainability. 2019;11(3):638.
- 10.
Lynch K. Good city form. Massachusetts Institute of Technology; 1981. 514 p.
- 11.
Lynch K. The image of the city; 1960. 194 p.
- 12.
Gehl J. Life between buildings; 2012.
- 13. Montgomery J. Making a city: urbanity, vitality and urban design. J Urban Des. 1998;3(1):93–116.
- 14. Sung H, Lee S. Residential built environment and walking activity: empirical evidence of Jane Jacobs’ urban vitality. Transp Res Part D: Trans Environ. 2015;41:318–29.
- 15. Amiruzzaman M, Curtis A, Zhao Y, Jamonnak S, Ye X. Classifying crime places by neighborhood visual appearance and police geonarratives: a machine learning approach. J Comput Soc Sci. 2021;4(2):813–37. pmid:33718652
- 16. Filion P, Hammond K. Neighbourhood land use and performance: the evolution of neighbourhood morphology over the 20th century. Environ Plann B Plann Des. 2003;30(2):271–96.
- 17. Zarin SZ, Niroomand M, Heidari AA. Physical and social aspects of vitality case study: traditional street and modern street in Tehran. Procedia Soc Behav Sci. 2015;170:659–68.
- 18. Wu J, Ta N, Song Y, Lin J, Chai Y. Urban form breeds neighborhood vibrancy: a case study using a GPS-based activity survey in suburban Beijing. Cities. 2018;74:100–8.
- 19. Liu H, Gou P, Xiong J. Vital triangle: a new concept to evaluate urban vitality. Comput Environ Urban Syst. 2022;98:101886.
- 20. Jia C, Liu Y, Du Y, Huang J, Fei T. Evaluation of urban vibrancy and its relationship with the economic landscape: a case study of Beijing. IJGI. 2021;10(2):72.
- 21. Kim Y-L. Seoul’s Wi-Fi hotspots: Wi-Fi access points as an indicator of urban vitality. Comput Environ Urban Syst. 2018;72:13–24.
- 22. Yue H, Zhu X. Exploring the relationship between urban vitality and street centrality based on social network review data in Wuhan, China. Sustainability. 2019;11(16):4356.
- 23. Zhang L, Chiaradia A. Three-dimensional spatial network analysis and its application in a high density city area, Central Hong Kong. UPI. 2019;34(1):46–53.
- 24. Feng D, Tu L, Sun Z. Research on population spatiotemporal aggregation characteristics of a small city: a case study on shehong county based on baidu heat maps. Sustainability. 2019;11(22):6276.
- 25. Tang J, Long Y. Measuring visual quality of street space and its temporal variation: methodology and its application in the Hutong area in Beijing. Landsc Urban Plann. 2019;191:103436.
- 26.
Wang X. Study on the sustainability of new urban districts and measuring in dimension of urban morphology in the Yangtze River Delta. Beijing: China Social Press; 2022.
- 27. Cervero R, Kockelman K. Travel demand and the 3Ds: density, diversity, and design. Transp Res Part D: Trans Environ. 1997;2(3):199–219.
- 28. Ewing R, Cervero R. Travel and the built environment. J Am Plann Assoc. 2010;76(3):265–94.
- 29. Gómez-Varo I, Delclòs-Alió X, Miralles-Guasch C. Jane Jacobs reloaded: a contemporary operationalization of urban vitality in a district in Barcelona. Cities. 2022;123:103565.
- 30. Li X, Li Y, Jia T, Zhou L, Hijazi IH. The six dimensions of built environment on urban vitality: fusion evidence from multi-source data. Cities. 2022;121:103482.
- 31. Li Q, Cui C, Liu F, Wu Q, Run Y, Han Z. Multidimensional urban vitality on streets: spatial patterns and influence factor identification using multisource urban data. IJGI. 2021;11(1):2.
- 32. Ge Y, Gan Q, Ma Y, Guo Y, Chen S, Wang Y. Spatial vitality detection and evaluation in Zhengzhou’s main urban area. Buildings. 2024;14(11):3648.
- 33. Xia C, Zhang A, Yeh AGO. The varying relationships between multidimensional urban form and urban vitality in Chinese megacities: insights from a comparative analysis. Ann Am Assoc Geogr. 2021;112(1):141–66.
- 34. Jiang Y, Huang Z, Zhou X, Chen X. Evaluating the impact of urban morphology on urban vitality: an exploratory study using big geo-data. Int J Digit Earth. 2024;17(1).
- 35. Jin A, Ge Y, Zhang S. Spatial characteristics of multidimensional urban vitality and its impact mechanisms by the built environment. Land. 2024;13(7):991.
- 36. Ye Y, Zeng W, Shen Q, Zhang X, Lu Y. The visual quality of streets: a human-centred continuous measurement based on machine learning algorithms and street view images. Environ Plan B: Urban Anal City Sci. 2019;46(8):1439–57.
- 37. Chen L, Lu Y, Sheng Q, Ye Y, Wang R, Liu Y. Estimating pedestrian volume using Street View images: a large-scale validation test. Comput Environ Urban Syst. 2020;81:101481.
- 38. Gong F-Y, Zeng Z-C, Zhang F, Li X, Ng E, Norford LK. Mapping sky, tree, and building view factors of street canyons in a high-density urban environment. Build Environ. 2018;134:155–67.
- 39. McLeod S. Maslow’s hierarchy of needs. Simply Psychol. 2007;1:1–18.
- 40. Wu C, Zhao M, Ye Y. Measuring urban nighttime vitality and its relationship with urban spatial structure: a data-driven approach. Environ Plan B: Urban Anal City Sci. 2022;50(1):130–45.
- 41. Jia J, Zhang X, Huang C, Luan H. Multiscale analysis of human social sensing of urban appearance and its effects on house price appreciation in Wuhan, China. Sustain Cities Soc. 2022;81:103844.
- 42. Dubey A, Naik N, Parikh D, Raskar R, Hidalgo CA. Deep learning the city: quantifying urban perception at a global scale. Comput Vis ECCV. 2016;9905:196–212.
- 43. He J, Zhang J, Yao Y, Li X. Extracting human perceptions from street view images for better assessing urban renewal potential. Cities. 2023;134:104189.
- 44. Ye Y, Richards D, Lu Y, Song X, Zhuang Y, Zeng W, et al. Measuring daily accessed street greenery: a human-scale approach for informing better urban planning practices. Landsc Urban Plann. 2019;191:103434.
- 45.
Li X, Ratti C. Using Google Street View for street-level urban form analysis, a case study in Cambridge, Massachusetts. In: The mathematics of urban morphology; 2019. p. 457–70.
- 46. Qi Y, Chodron Drolma S, Zhang X, Liang J, Jiang H, Xu J, et al. An investigation of the visual features of urban street vitality using a convolutional neural network. Geo-spat Inf Sci. 2020;23(4):341–51.
- 47. Ogawa Y, Oki T, Zhao C, Sekimoto Y, Shimizu C. Evaluating the subjective perceptions of streetscapes using street-view images. Landsc Urban Plan. 2024;247:105073.
- 48. Wu C, Peng N, Ma X, Li S, Rao J. Assessing multiscale visual appearance characteristics of neighbourhoods using geographically weighted principal component analysis in Shenzhen, China. Comput Environ Urban Syst. 2020;84:101547.
- 49. Dongfeng Y, Xiaomeng W, Ruina H. Nonlinear and synergistic effects of the built environment on street vitality: The case of Shenyang. Urban Plan Forum. 2023;(5):93–102.
- 50. Marukatat S. Tutorial on PCA and approximate PCA and approximate kernel PCA. Artif Intell Rev. 2022;56(6):5445–77.
- 51. Wang C, Wang B, Wang Q, Lei Y. Nonlinear associations between urban vitality and built environment factors and threshold effects: a case study of central Guangzhou City. Prog Geogr. 2023;42(1):79–88.
- 52. Wenhao Y, Tinghua A. The visualization and analysis of POI features under network space supported by kernel density estimation. Acta Geod Cartogr Sin. 2015;44(1):82–90.
- 53. Wang J, Li YG, Meng B, Wu J, Bo Y, et al. Spatiotemporal data analysis in geography. Acta Geogr Sin. 2014;69(9):1326–45.
- 54. Woodard JD, Schnitkey GD, Sherrick BJ, Lozano‐Gracia N, Anselin L. A spatial econometric analysis of loss experience in the U.S. Crop Insurance Program. J Risk Insur. 2010;79(1):261–86.
- 55. Jingyi C, Hongxia L, Linli H, Qiuhong L, Wenjing J. Correlation between social economic development and ecosystem service value——A case of Chongqing metropolitan area. Res Soil Water Conserv. 2017;24(04):330–6.
- 56. Winham SJ, Colby CL, Freimuth RR, Wang X, de Andrade M, Huebner M, et al. SNP interaction detection with Random Forests in high-dimensional genetic data. BMC Bioinform. 2012;13:164. pmid:22793366
- 57. Biau G. Analysis of a random forests model. J Mach Learn Res. 2012;13:1063–95.
- 58. Breiman L. Random forests. Mach Learn. 2001;45(1):5–32.
- 59. Hansen LK, Salamon P. Neural network ensembles. IEEE Trans Pattern Anal Machine Intell. 1990;12(10):993–1001.
- 60. Li Y-Z, Li Z-J, Zheng S-W, Lin J-K, Yang S-Y, Zhao L, et al. Elucidating the spatial differentiation of soil organic matter and influencing factors within the Yihe River Basin. Huan Jing Ke Xue. 2025;46(6):3865–76. pmid:40582821
- 61. Anselin L, Griffith DA. Do spatial effecfs really matter in regression analysis? Pap Reg Sci. 1988;65(1):11–34.
- 62. Zhang F, Zu J, Hu M, Zhu D, Kang Y, Gao S, et al. Uncovering inconspicuous places using social media check-ins and street view images. Comput Environ Urban Syst. 2020;81:101478.
- 63. Liu F, Tang Z, Zhang L, Yu L, Ke L. Exploring the relationship between spatiotemporal distribution of urban vibrancy and neighborhood attributes by coupling multi-source data: a case of Nanshan District in Shenzhen. Trop Geogr. 2025;45(03):410–22.
- 64. Sun H, Yupei J. Spatial heterogeneity of the impact of built environment on urban vitality: a case study of the central urban area of Nanjing. Geogr Res. 2024;43(07):1700–14.
- 65. Ning Z, Zhibin Z, Qiangian G, Xiaomin M, Nana Z, Huaying L. Spatial pattern of urban vitality of national new districts in inland northwest China. J Arid Land Resour Environ. 2024;38(04):38–49.
- 66. Friedmann J. Where we stand: a decade of world city research. In: World cities in a world-system. Cambridge University Press; 1995. p. 21–47.
- 67. Newman P, Kenworthy J. Sustainability and cities: overcoming automobile dependence. Sch Archit Harbin Instit Technol. 1999;44(4):219–26.
- 68. Tang L, Lin Y, Li S, Li S, Li J, Ren F, et al. Exploring the influence of urban form on urban vibrancy in shenzhen based on mobile phone data. Sustainability. 2018;10(12):4565.
- 69. Ye Y, Yeh A, Zhuang Y, van Nes A, Liu J. “Form Syntax” as a contribution to geodesign: a morphological tool for urbanity-making in urban design. Urban Des Int. 2016;22(1):73–90.
- 70. Shi S, Gou Z, Chen LHC. How does enclosure influence environmental preferences? A cognitive study on urban public open spaces in Hong Kong. Sustain Cities Soc. 2014;13:148–56.