Evaluation of comprehensive vitality of Shanghai’s commercial centers using multi-dimensional geospatial big data

Abstract

Understanding the vitality of commercial centers is essential for urban planning and economic development. This study introduces a novel framework to quantitatively assess the vitality of commercial centers in Shanghai, leveraging multi-source geospatial big data, including the rarely utilized UnionPay consumption data, alongside mobile signaling data and Points of Interest (POI). By integrating these datasets, a multi-dimensional vitality index was developed, encompassing population, goods, and market dynamics. Dimensionality reduction techniques such as Principal Component Analysis and Factor Analysis were applied, followed by K-means clustering to classify commercial centers into vitality clusters. Unlike prior studies which relied primarily on POI and mobility data, our inclusion of granular consumption metrics provides a deeper understanding of market vitality. Results reveal a spatial gradient, with higher vitality concentrated in central Shanghai and diminishing towards the periphery, validating the “commercial gravity” theory. This framework not only enhances the methodological toolkit for evaluating urban commercial vitality but also offers valuable insights for optimizing commercial resources and addressing spatial inequalities in urban planning. The approach is adaptable for other metropolitan regions.

Introduction

Urban vitality is a cornerstone of sustainable urban development and significantly influences the quality of life for city dwellers¹. Vibrant cities foster comfortable living environments and fulfill diverse social and economic needs, benefiting residents and businesses alike². As a critical component of urban vitality, the dynamism of commercial centers reflects the intensity of economic activities, consumer engagement, and service provision³. Modern commercial centers offer multifaceted functions—ranging from shopping and dining to entertainment and leisure—making their vitality a barometer of urban economic health. Among these factors, consumption serves as a direct and indispensable measure of commercial vitality, capturing the economic transactions and consumer preferences that drive urban growth⁴.

Research on the attractiveness and vitality of commercial centers has garnered increasing attention from governments, businesses, and academia. However, the quantitative evaluation of comprehensive commercial vitality at a city scale remains underexplored⁵. Existing studies often focus on isolated metrics, such as pedestrian flow or POI density, neglecting the interplay of population dynamics, market performance, and consumption behaviors^6,7. For example, mobile signaling data alone only captures "where people are" but not "what they do". Additionally, traditional approaches lack the granularity required for actionable insights, often overlooking spatial disparities that could guide effective urban planning⁸. The classical retail gravity models proposed by Reilly⁹ and Huff¹⁰ underscore the importance of size and accessibility in determining the attractiveness of commercial centers, neglecting how population mobility interacts with consumption patterns. While these models laid the theoretical foundation for analyzing commercial dynamics, contemporary urban environments demand more sophisticated methodologies to capture the complexity of modern consumer behavior and resource distribution^11,12.

Shanghai, as China’s economic and financial epicenter, presents a compelling case for studying commercial vitality. Its ongoing urbanization and strategic initiatives, such as the "Shanghai Sectoral Planning of Commercial Space (2022–2035)," aim to enhance the city’s role as an international consumption hub¹³. This study introduces a comprehensive framework to evaluate the vitality of Shanghai’s commercial centers using multi-source geospatial big data. By integrating rarely used UnionPay consumption data with mobile signaling and POI datasets, we propose a robust methodology to assess population, goods, and market dynamics. Our approach not only enriches theoretical understanding but also provides actionable insights for optimizing commercial planning and addressing spatial inequalities.

The remainder of this article is structured as follows: "Literature review" section reviews the relevant literature, highlighting gaps addressed by this study. "Materials and methods" section outlines the proposed framework and methodology. "Results" section presents the results, including spatial patterns and clustering of commercial vitality. "Discussion" section discusses the findings’ implications for theory and practice, while "Conclusion" section concludes with recommendations for future research and applications.

Literature review

Concept of commercial vitality

Urban vitality is a multifaceted concept that integrates economic, social, and cultural dynamics. Commercial centers, as key urban nodes, exemplify this vitality through their ability to attract people, provide services, and drive economic transactions¹⁴. The concept of commercial vitality has been categorized into three key elements: place (commercial centers as physical settings), activity (economic transactions and foot traffic), and user (the population engaging with these spaces)^15,16.

Scholars have assessed commercial vitality from two main perspectives. The site function perspective focuses on macro- and meso-level attributes, such as the comparative attractiveness of commercial centers based on indicators like size, accessibility, and service diversity^3,17. In contrast, the population perspective examines spatial and temporal variations in user behaviors, leveraging tools such as GPS tracking and mobile phone signaling to capture movement patterns within commercial areas^14,18.

Building on Jacobs¹⁶ and Xia et al.⁴, we define commercial vitality as the synergistic capacity of urban spaces to sustain high-intensity economic transactions, population flows, and service provision, measured through multidimensional spatiotemporal interactions. Indicators such as shopping services, retail diversity, and market density are critical proxies for evaluating commercial vitality^4,19. This paper builds on these foundations by integrating population, goods consumption, and market dynamics into a multi-dimensional evaluation framework.

Factors influencing commercial vitality

Commercial vitality is influenced by a range of demographic, economic, and infrastructural factors. Early theories, such as Reilly’s⁹ “retail gravity” model and Huff’s¹⁰ probabilistic trade area analysis, emphasized the role of size and proximity in determining commercial attractiveness. Contemporary studies have expanded on these ideas by incorporating diverse urban characteristics.

Population density, diversity, and mobility remain critical determinants of vitality. Jacobs²⁰ emphasized that vibrant urban spaces thrive on dense, diverse populations, while Kim²¹ and Gómez-Varo et al.²² demonstrated how floating populations and dynamic movement patterns can enhance urban vibrancy. Retail infrastructure and economic activities also play key roles. Mixed land use, retail clustering, and competitive pricing have been linked to higher vitality²³.

Despite the breadth of research, studies often neglect dynamic consumption metrics, which provide direct insights into economic vitality. Prior works^7,14 relied on POI density and pedestrian flow as proxies for commercial vitality, which cannot distinguish between physical infrastructure presence (POIs) and actual economic transactions (consumptions). For example, a shopping mall with low foot traffic may appear “vibrant” in POI-based models but is economically inactive. Consumption behaviors, captured through credit card transactions or similar data, offer an untapped avenue for understanding urban commercial dynamics. This study addresses this gap by incorporating UnionPay consumption data, providing a novel lens to evaluate economic activity and its spatial distribution.

Application of geospatial data

The advent of geospatial big data has revolutionized urban vitality research. Traditional approaches, such as surveys or manual observations, have been supplanted by datasets offering real-time, granular insights into urban dynamics. Point of Interest (POI) data, mobile phone signaling, and nighttime light intensity have become standard tools for evaluating vitality^24,25.

For example, studies by Zhang et al.¹⁴ in Shanghai and Tang and Ta²⁶ in Central China leveraged POI data and heatmaps to explore the spatial–temporal patterns of urban vitality. Similarly, mobile phone data has been used to monitor population flows, providing a robust indicator of activity levels within commercial centers^27,28. While these approaches capture critical dimensions of vitality, they often overlook consumption behaviors.

This study uniquely incorporates UnionPay transaction data to evaluate consumption alongside spatial patterns. By combining consumption metrics with POI and mobility data, the proposed framework offers a holistic view of commercial vitality, bridging methodological gaps in current research.

Materials and methods

Research area

Shanghai, China’s largest economic and financial hub, serves as the research area for this study. Covering an area of 6340 km² and home to 24.88 million residents as of 2020, the city provides a dynamic context for examining commercial vitality. Renowned for its well-developed retail network, Shanghai’s tertiary sector contributes 73.1% of its GDP, supported by a robust consumer market and high per capita disposable income. The "Shanghai Sectoral Planning of Commercial Space (2022–2035)" initiative aims to enhance international consumption zones and stimulate vitality across urban and suburban areas, including the "Five New Cities" initiative¹³. The plan defines two levels of 34 commercial centers—20 municipal-level and 14 regional-level—distributed across 15 districts (Fig. 1, Table 2).

Fig. 1

Municipal and regional commercial centers in Shanghai (the base map was provided by Esri, using ArcGIS Online Services by ArcGIS Pro 3.4, https://pro.arcgis.com).

Most of the 22 municipal-level commercial centers are located within the central city, an area of approximately 667 km² enclosed by Shanghai’s outer ring road. This includes People’s Square, the city’s most prominent commercial hub (Fig. 1). For this study, the vitality of all 34 commercial centers identified in Shanghai’s commercial space planning was evaluated using a comprehensive vitality assessment framework (Table 1). To further illustrate vitality indices, the study focuses specifically on the 18 municipal-level centers within the outer ring road.

Table 1 detail information of commercial centers.

Vitality framework

This research proposes a synthesized framework for the comprehensive evaluation of commercial center vitality through a structured process involving indices, modeling, and application (Fig. 2). First, indices for three dimensions of vitality are established based on categorized features: Population vitality is rooted in the population perspective, analyzing dynamic human interactions through mobile signaling data¹⁸. Goods vitality and market vitality derive from the site function perspective, evaluating commercial infrastructure and transactional patterns via POI and UnionPay data³. This tripartite framework bridges the population-site duality, advancing beyond classical models^9,10.

Fig. 2

Indexes framework of commercial vitality evaluation process.

Second, in the modeling process, all features are spatially analyzed within the defined boundaries of each commercial center. The data is then visualized using a GIS platform to map spatial distributions and calculate statistics, facilitating the identification of distribution patterns across vitality categories. The first step in the analysis was to apply Principal Component Analysis (PCA) to reduce the dimensionality of the data and extract the most significant components. Next, Factor Analysis (FA) was used to identify latent factors underlying the principal components and generate standardized vitality scores. Finally, the resulting vitality scores were used as input for K-means clustering, which grouped the commercial centers into clusters based on their vitality characteristics. This workflow systematically reveals the multi-layered characteristics of Shanghai’s commercial vitality.

Ultimately, the comprehensive vitality analysis supports both the evaluation of commercial centers and urban commercial planning. The framework, grounded in the population-goods-market concept, utilizes eight core indicators further defined into 21 sub-indices.

The 21 sub-indices were derived through a synthesis of established urban vitality theories and Shanghai’s unique commercial context³. Population Vitality leverages mobile signaling data to quantify total population and distinguish between work population and residential population, capturing dynamic human interactions across spatiotemporal dimensions^2,18. Goods Vitality utilizes UnionPay consumption data to measure total consumption (amount and frequency) and category-specific expenditures (catering, retail, and accommodation), directly quantifying economic activity and addressing the "supply–demand decoupling" in traditional POI-based studies^4,15. Market Vitality integrates POI-based metrics, including neighboring features (number of residential blocks and households), commercial offices (number of office buildings, total office area, and floors), commercial facilities (number of shopping centers, supermarkets, and convenience stores), and business services (number of restaurants, hospitals, education facilities, banks, and transportation sites). This multidimensional framework bridges theoretical rigor with empirical relevance, ensuring robust vitality assessment. Specific details of these features are provided in Table 2.

Table 2 Vitality indicators and feature types.

Data

The data used in this study includes city administrative boundaries, mobile signaling-based population data, UnionPay consumption data, and Points of Interest (POIs) such as shopping malls and transportation sites. All attribute information related to population, goods, and market vitality is collected within the spatial boundaries of each commercial center and processed in GIS software as distinct geospatial layers. To better quantify differences among commercial centers, vitality indices are calculated per unit area (per km²), normalizing for variations in the size of each center. The GIS platform employs vector layers represented as points (POIs) and polygons (commercial centers). All data layers are preprocessed in Esri shapefile format and projected into a uniform coordinate system (e.g., EPSG: 3857, Web-Mercator).

Mobile signaling based population

The mobile signaling-based population data, collected for the year 2021, provides high-resolution information with a grid size of 50 m. This granularity enables meso-scale and micro-scale visualization and analysis in a GIS environment, offering valuable insights into the spatial–temporal distribution of various population groups. The fixed population is categorized into two types: (1) Work Population: Represents individuals active during working hours on weekdays, primarily reflecting the workforce concentrated in commercial centers. (2) Residential Population: Represents individuals staying overnight on weekdays and active during daytime hours on weekends, capturing the residential aspect of population vitality.

The total population is calculated by eliminating duplications between the work and residential populations, providing an accurate measure of the overall population distribution within each commercial center (Table 2). This data supports detailed analysis of activity patterns, offering a foundation for understanding how commercial centers function as hubs of interaction.

UnionPay consumption

UnionPay consumption big data captures detailed consumption information within the spatial boundaries of Shanghai’s commercial centers for the year 2021. The indicators include total consumption and categorized consumption metrics, such as consumption amount, frequency of transactions, and expenditures on catering, retail, and accommodation (Table 2). The UnionPay consumption dataset, which includes monthly average expenditure and transaction counts for each commercial center, underwent rigorous cleaning to ensure consistency and reliability. Each commercial center was treated as an independent sample, and no missing values were present in the dataset. To address potential outliers, extreme values were identified and removed using a ± 3σ threshold, ensuring data consistency without compromising the overall distribution. This preprocessing step guarantees that the analysis is based on robust and representative data, minimizing the impact of anomalies on the results. This dataset has been validated against city statistical data, ensuring its reliability and accuracy. It complements the statistical yearbook by providing granular insights into consumption patterns across varying scales.

At the macro-scale, the data reflects the total consumption volume of the city over a defined period, offering a comprehensive view of urban economic activity. At the meso-scale, it enables analysis of consumption categories, revealing specific characteristics and comparative advantages of individual commercial centers. This information is instrumental in understanding the economic dynamics of Shanghai’s commercial hubs and serves as a robust supplementary source for urban economic studies.

Features of POIs

POI data was accessed through the AutoNavi platform, providing detailed information on various categories, including shopping services, catering services, and daily-life services. For this study, subcategories with significant proportions in municipal commercial centers were selected to represent the commercial format structure of these centers. To analyze the traffic convenience of commercial centers, transportation-related facilities such as bus stations, subway stations, railway stations, and airports were included. Commercial facilities were represented by the number of shopping centers, convenience stores, and supermarkets, while business services were assessed using the number of establishments offering catering, retail, education, banking, and healthcare services (Table 2). All POI data were collected for the year 2021 and validated through field surveys to ensure accuracy and consistency.

Methods

Statistical analysis

Principal component analysis (PCA) reduces dimensionality by transforming correlated variables into uncorrelated principal components, preserving key features of the original data. This study used PCA to simplify the multi-dimensional vitality indices by identifying major contributing components.

Factor analysis (FA), an extension of PCA, investigates correlations among variables to identify latent representative factors. FA aims to uncover hidden structures within the data by determining whether a group of variables can act as indicators of underlying, unobservable factors. This approach is particularly useful for verifying hypotheses regarding variable relationships and exploring the structure of the data.

In this study, FA was implemented following PCA to analyze the internal dependencies among multi-dimensional variables and further extract common factors from variable groups. The process examines the correlation coefficient matrix to identify a few unobservable dominant variables that represent the original dataset comprehensively. These extracted factors provide a simplified yet accurate representation of the data’s internal structure.

The comprehensive vitality score was calculated by weighting the principal component scores according to their variance contributions. This approach ensures that the indices reflect the most significant dimensions of the data. The calculation follows Eq. (1):

$$V_{i} = \sum\limits_{{k = 1}}^{n} {W_{k} } F_{{ik}}$$

(1)

where ${V}_{i}$ is the k-th factor score for the i-th center, and ${W}_{k}$ is the corresponding weight.

Spatial clustering

Clustering is an unsupervised learning algorithm used to group data into categories that maximize internal similarity within groups and minimize external similarity between groups. The K-means clustering algorithm is widely used due to its speed, parallel processing capabilities, and strong performance in discriminating spatially distributed groups. The algorithm divides n points into k clusters, ensuring that each point belongs to the cluster with the closest mean center. The calculation follows Eq. (2):

$$arg\,\underset{s}{\mathit{min}}\sum_{i=1}^{k}\sum_{x\in {s}_{i}}{\Vert x- {\mu }_{i}\Vert }^{2}$$

(2)

where k is the number of clusters of the class, and μ is the mean of all points in S_i. and the algorithm iteratively classifies objects based on their proximity to cluster centers.

To ensure robust clustering results, we performed multiple iterations of K-means clustering using the Multivariate Clustering tool in ArcGIS Pro. The comprehensive vitality index and individual dimensions of "people-goods-market" vitality were used as input variables. The algorithm sought to maximize within-group homogeneity and between-group distinctiveness.

To evaluate the accuracy of classification, R² score is selected to compare the performance among many tries of cluster numbers, with different combinations of vitality indexes, as described by following Eqs. (3):

$${R}^{2}\text{=}\text{1} - \frac{\sum_{i=1}^{n}{({\widehat{y}}_{i}-{y}_{i})}^{2}}{\sum_{i=1}^{n}{({y}_{i}-\overline{y })}^{2}}$$

(3)

where n indicates the numbers of samples are in the dataset, ${\widehat{{\varvec{y}}}}_{{\varvec{i}}}$ represents the i-th sample’s predicted value, and ${{\varvec{y}}}_{{\varvec{i}}}$ is the i-th sample’s true value.

Results

Vitality of population

The spatial distribution of population vitality reveals a clear gradient, with higher values concentrated in central urban commercial centers and progressively lower values towards the periphery. Among the 18 municipal-level commercial centers, East Nanjing Road (ENJ), Lujiazui (LJZ), and West Nanjing Road (WNJ) exhibit the highest total population, exceeding 200,000 individuals (Fig. 3a). This is primarily attributed to their strategic locations within Shanghai’s central business district and their proximity to key transportation hubs. In contrast, developing commercial centers such as Disney Resort (DR) and Nanhui (NH) show significantly lower population vitality, with total populations below 50,000. Additionally, the analysis highlights a predominance of working populations over residential populations across most commercial centers, except for Zhenru (ZR), where residential populations slightly surpass working populations.

Fig. 3

Distribution of vitalities of commercial centers: population vitality (a), and consumption vitality (b). (Created by ArcGIS Pro 3.4, https://pro.arcgis.com).

Vitality of goods

Goods vitality demonstrates a strong positive correlation with centrality and economic density. The highest consumption levels are observed in ENJ, LJZ, and WNJ, where total transaction volumes exceed USD 5 billion (Fig. 3b). These centers are characterized by their dense retail landscapes and high consumer traffic. The composition of consumption further highlights retail as the dominant category across all commercial centers, accounting for over 50% of total spending in most locations. Notable exceptions include Disney Resort (DR), where catering contributes a higher proportion, and North Bund (NB), which has the highest percentage of accommodation-related expenditures. Peripheral centers generally exhibit lower goods vitality, reflecting limited retail and service infrastructure.

Vitality of markets

Market vitality is assessed through four sub-indicators: neighboring facilities, commercial offices, commercial facilities, and business services. The results indicate significant variation in these indicators among the 18 municipal-level centers:

Neighboring facilities

ENJ, LJZ, and WNJ have the highest density of residential communities and households, reinforcing their roles as urban hubs (Fig. 4a). Peripheral centers like DR and NH show limited neighboring facilities, hindering their potential as vibrant commercial nodes.

Fig. 4

Distribution of market vitality of commercial centers: neighboring (a), commercial office (b), commercial facilities (c), and business services (d). (Created by ArcGIS Pro 3.4, https://pro.arcgis.com).

Commercial offices

Central centers such as ENJ and LJZ exhibit the highest concentration of office buildings, with accumulative floor areas exceeding 1 million square meters (Fig. 4b). This contrasts with suburban centers, where office infrastructure is sparse.

Commercial facilities

ENJ and WNJ lead in the number of shopping centers, supermarkets, and convenience stores, highlighting their comprehensive retail ecosystems (Fig. 4c). DR and NH remain underdeveloped, with fewer than 50 such facilities.

Business services

Catering and banking services dominate the business landscape in most commercial centers. For example, ENJ hosts over 1000 catering establishments, while LJZ leads in banking and financial services (Fig. 4d).

In all, the commercial centers in the central city area, such as WNJ, ENJ, LJZ, MHH, YY, and XJH have obvious commercial markets advantages compared to other commercial centers, showing a spatially clustering pattern of advantages in terms of neighboring, commercial office, commercial facility, and business services. Different commercial centers have different advance type of fields, reflecting the regional attributes and resource supply capabilities of each commercial center.

Comprehensive vitality evaluation

Statistical analysis

The initial factor loading matrix underwent variance maximization rotation, resulting in eigenvalues greater than 1 for all three principal components. These components achieved a cumulative variance contribution rate of 76.9%. The measured variables for the indicators of market vitality, population vitality, and goods vitality were rotated through principal component analysis, each yielding a single principal component.

The cumulative explained variance for the principal components was 72.43%, 79.12%, 77.03%, 71.01%, 97.44%, and 85.39% across the variables (see Table 3). This indicates that the single principal component extracted for the four market vitality indicators—neighboring facilities, commercial offices, commercial facilities, and business services—adequately represents the original multiple variables. These results confirm that the dimensionality reduction process effectively captures the primary characteristics of the data.

Table 3 Statistical information of PCA.

Vitality index and ranking

The results of PCA indicate that the cumulative explained variance of the principal component is 76.9%, confirming that the extracted principal component effectively represents the overall vitality characteristics of commercial centers. The comprehensive scores for evaluating vitality were calculated using the principal component equation (Equation FA) and ranked in ascending order to determine the vitality levels of commercial centers (Table 4).

Table 4 Comprehensive value of commercial vitality of all commercial centers.

The analysis reveals that the three municipal business districts—East Nanjing Road, Lujiazui, and West Nanjing Road—achieved the highest factor scores, securing the top three positions in the comprehensive vitality ranking. These are followed by Huaihai Middle Road, North Sichuan Road, Xujiahui, Yuyuan Garden, Zhongshan Park, Hongqiao Gubei, and Wujiaochang, ranked fourth to tenth. The remaining municipal business districts scored below 0 and are ranked from 11 to 34th (Table 4).

To enable more intuitive comparisons, the scores were normalized to a range of 0 to 10, producing the comprehensive vitality values for each commercial center, as shown in Fig. 5. The spatial distribution pattern reveals a significant clustering of high-vitality commercial centers in the central region, predominantly within the outer ring road. In contrast, commercial centers located along the city outskirts exhibit relatively low vitality values, with southern regions—e.g., Songjiang (SJ), Xinzhuang (XZ), Pudong Jinqiao (PJ), Disney Resort (DR), Fengxian (FX), Sheshan (ST), and Nanhui (NH)—ranking notably lower than northern regions such as Qingpu (QP), Hongqiao (HQ), Jiading (JD), and Qibao (QB). This disparity is largely attributed to the greater distance of southern regions from the city center compared to their northern counterparts.

Fig. 5

Spatial distribution of comprehensive vitality value of commercial centers (Created by ArcGIS Pro 3.4, https://pro.arcgis.com).

K-means clusters

The results of K-means clustering reveal four distinct clusters based on the comprehensive vitality index, with an R² score of 94.5%, indicating a high degree of classification accuracy. As shown in Fig. 6, Cluster 1 comprises commercial centers with the highest vitality due to a combination of high consumer spending, dense retail environments, and high population density. These centers are major commercial hubs with diverse consumer activities and strong economic infrastructure. This cluster includes East Nanjing Road (ENJ), Lujiazui (LJZ), and West Nanjing Road (WNJ), all located within the international consumption zone of the city center.

Fig. 6

Four clusters of commercial centers classified by comprehensive vitality value index (Created by ArcGIS Pro 3.4, https://pro.arcgis.com).

Cluster 2 includes North Sichuan Road (NSC), Xujiahui (XJH), North Bund (NB), and Yu Garden (YY), shows relative-high vitality, driven by strong retail presence and population but lacking in other infrastructural elements such as office space and public services. Notably, the commercial centers in Clusters 1 and 2 align closely with those identified as part of the International Consumption Agglomeration Zone in the Shanghai Sectoral Planning of Commercial Space (2022–2035)¹³.

Cluster 3 and Cluster 4 represent areas with lower vitality, which are characterized by limited consumer traffic, lower spending, and fewer commercial and infrastructural facilities. Cluster 3 primarily comprises municipal commercial centers, with 7 out of 11 located within the outer ring road. Most centers in this cluster are situated in the northern part of Shanghai, with the exception of Binhai (BH), which lies outside the city core.

In contrast, Cluster 4 includes commercial centers distributed equally inside and outside the outer ring road, with a majority located in the southern regions of the city. This spatial distribution underscores a notable north–south divide in commercial vitality, with northern centers generally exhibiting higher vitality values than their southern counterparts.

Discussion

Key findings

This study leverages multi-source geospatial big data, including POIs, mobile signaling, and consumption metrics, to construct a comprehensive business vitality framework. By integrating statistical models, such as PCA for dimensionality reduction, FA for structural equation modeling, and K-means clustering for spatial agglomeration analysis, the framework enables a quantitative evaluation of vitality and effectively identifies the spatial patterns of commercial centers.

Across municipal commercial centers, the working population consistently exceeds the resident population (Fig. 3a). Commercial centers with high consumption levels tend to have larger populations and exhibit notable resource supply advantages. Retail businesses account for the largest proportion of consumption, as seen in Fig. 3b. Convenience stores (Fig. 4c) represent the largest share of commercial facilities, while catering services (Fig. 4d) dominate business services. All vitality categories, including population, consumption, neighboring facilities, commercial offices, and business services, exhibit a significant spatial gradient, decreasing from the central urban areas to the periphery (Figs. 3 and 4).

The central commercial centers within Shanghai’s inner ring road demonstrate significant comparative advantages²⁹. Notably, despite the high concentration of facilities in ENJ, WNJ, and LJZ, their vitality components vary substantially: ENJ excels in catering; XJH leads in education; LJZ dominates in banking and finance; WNJ is prominent in healthcare. These findings highlight the importance of quantitatively evaluating comprehensive vitality to capture inter-center variations³⁰.

The PCA and FA model explains 76.9% of the variance, confirming its reliability in representing comprehensive vitality (Table 3). ENJ achieves the highest scores in goods vitality (1.84) and population vitality (2.86), resulting in the highest overall vitality ranking. Vitality scores vary significantly among the top-ranked centers (ranks 2 to 10), indicating that high comprehensive vitality does not guarantee uniform advantages across all categories (Table 4).

The spatial distribution of comprehensive vitality shows a predictable decline from the city center to the outskirts, with pronounced agglomeration effects in central areas (Fig. 5). K-means clustering identifies Clusters 1 and 2 as having the most centralized and resource-rich commercial centers. These clusters align with the "International Consumption Agglomeration Zone" outlined in the Shanghai Sectoral Planning of Commercial Space (2022–2035) (Fig. 6).

Theoretical implications

Classical retail theories, such as Reilly’s "law of retail gravitation"⁹ and Huff’s probabilistic model¹⁰, suggest that the attractiveness of a commercial center is directly proportional to its size and inversely proportional to the time and distance required to access it^31,32. Our findings reinforce the “commercial gravity” theory, demonstrating that the most vibrant commercial centers, with the highest comprehensive vitality scores, are densely concentrated within the central activity region of the city. The analysis further reveals a clear spatial boundary, distinguishing vitality patterns between northern and southern regions, as well as a correlation between vitality levels and proximity to the city center. For example, clusters 3 and 4 highlight the declining vitality in outskirts, influenced by increased distance from the urban core (Fig. 6).

Traditional index generation methods, such as the Analytic Hierarchy Process (AHP), often rely on subjective weighting, introducing biases and limiting applicability to quantitative models³³. Additionally, the persistent "snapshot vs. process" debate in urban vitality studies stems from the disconnect between static infrastructure metrics (e.g., POI) and dynamic economic behaviors (Garau & Annunziata, 2022). To overcome these limitations, we propose a novel "population-goods-market" framework that integrates static and dynamic measures (POI and consumption data), leveraging PCA and FA to objectively extract multidimensional insights. This data-driven approach enables robust vitality quantification while resolving theoretical and methodological gaps in traditional research.

This approach not only advances the theoretical understanding of retail and urban vitality but also provides practical implications for optimizing the attractiveness and resource allocation of business centers in mega-cities like Shanghai. By quantitatively evaluating comprehensive vitality, this framework contributes to the development of more equitable and efficient commercial planning strategies, enhancing both theoretical and practical insights into urban economic dynamics.

Application implications

The integration of multi-source big data is revolutionizing the strategic improvement of commercial center vitality in urban planning. Multi-dimensional geospatial data provides a unique advantage by presenting vitality metrics in a map format, enabling intuitive visualization for urban planners. This approach facilitates communication between planners and decision-makers, supporting efforts to monitor resource imbalances, identify shortages in specific vitality dimensions at the regional level, and optimize overall vitality across the city.

Comprehensive evaluations of commercial vitality offer valuable insights for the development of both municipal and regional commercial centers. For municipal centers, the analysis confirms that the core commercial hubs—represented by the top seven centers (Clusters 1 and 2)—demonstrate strong attractiveness, aligning with the municipal government’s international consumption zone strategy. However, the findings also highlight significant challenges for peripheral commercial centers, where greater distances from the city center hinder their ability to attract populations and stimulate consumption. Despite municipal efforts to enhance vitality in key-node centers, such as the "Five New Cities" (Qingpu, Songjiang, Jiading, Fengxian, and Nanhui) and developing international consumption zones like Disney Resort and Hongqiao, these centers still rank relatively low in comprehensive vitality.

Policymakers should prioritize investments in low-vitality areas by improving transportation connectivity, expanding retail services, and incentivizing businesses to invest in these areas. For instance, due to its distance from the city center, areas like NH and DR could benefit from enhanced retail infrastructure and better access to transportation hubs. To address this disparity, peripheral centers like Disney Resort (DR) shall prioritize transit-oriented development. Policies aimed at increasing population inflow and boosting commercial activity should be prioritized as part of the Nanhui New City Development Plan^34,35. Similarly, targeted strategies should focus on strengthening specific weaknesses in low-performing centers, ensuring alignment with their planning goals, such as enhancing the nighttime economy in worker-dominated centers (e.g., LJZ).

A limitation of this study is its use of aggregated data for the entire year, which overlooks temporal variations in commercial vitality. Future research could benefit from more granular temporal data, such as monthly or seasonal data, to capture dynamic changes. Furthermore, employing functional data analysis (PCA adapted for functional data) could provide more nuanced insights into the temporal patterns of urban vitality^36,37. Future applications could incorporate additional geospatial features into the evaluation framework. For instance, customer satisfaction ratings from platforms like TripAdvisor and consumption data at finer granularities (e.g., individual shopping malls) could enhance the precision of vitality assessments. Exploring the origins of consumption, whether from within or outside Shanghai, would also provide valuable insights into regional economic dynamics. Also, UnionPay data may not fully capture the behavior of certain demographic groups, particularly those who are less likely to use UnionPay, such as older adults or low-income populations. This limitation may affect the generalizability of the results, and future research could explore the use of alternative data sources to address this issue. Moreover, adopting advanced machine learning models could improve cluster optimization and further refine the analysis, representing a promising direction for urban planning and commercial industry applications.

Conclusion

Business centers attract significant attention in urban planning applications, yet quantitative assessments of their vitality remain underexplored. This study integrates multi-dimensional features—population flow, goods dynamics within business centers, and market consumption—to comprehensively evaluate the population-goods-market vitality of commercial centers in the commercial hub of Shanghai. The generation of comprehensive vitality scores is achieved through statistical models, including PCA and FA, enabling an analysis of the spatial distribution and clustering of both regional and municipal commercial centers. This comprehensive framework not only provides richer insights into urban vitality but also offers a more dynamic and granular perspective compared to traditional retail gravity models and probabilistic trade area analyses.

The results reveal a pronounced agglomeration of high vitality values in Shanghai’s city center, with the top seven commercial centers (out of 34) demonstrating strong alignment with the “commercial gravity” theory. These high-performing centers fall within the same clusters, which correspond spatially to the Shanghai Municipal Government’s development plan for the "International Consumption Zone." Furthermore, a clear decreasing trend in comprehensive vitality is observed as the distance from the city center increases. Notably, municipal commercial centers do not consistently exhibit higher vitality than regional commercial centers, particularly in recently developed and more distant locations, such as Hongqiao and Disney Resort.

These findings highlight the potential of applying multi-dimensional geospatial data and statistical models to advance methodologies for evaluating vitality in the commercial field. This work provides critical insights into the development levels of Shanghai’s major urban construction initiatives, such as the "Five New Cities" and the "Hongqiao International Central Business District," offering valuable guidance for municipal government strategies. For practical applications, the proposed framework facilitates quantitative evaluations of commercial centers and serves as a reference for strategic commercial planning in other cities (e.g., Tokyo, New York).