Classifying Metro Station Areas for Urban Regeneration: An RFM Model Approach and Differentiated Strategies in Beijing

Abstract

Amid growing demands for urban regeneration, metro station areas (MSAs) have emerged as critical spatial units for assessing renewal potential. However, their highly heterogeneous functional and spatial attributes pose challenges to precise classification and targeted strategy development. This study introduces the RFM (recency, frequency, and monetary) model—originally used in marketing—to the urban renewal domain. By mapping POI (point of interest) data, population density, and land price to the RFM dimensions, a three-dimensional evaluation framework is constructed. Using QGIS to process multi-source data for 118 MSAs in Beijing, we apply an improved five-quantile stratification method to classify station areas into eight renewal potential types. The results reveal a concentric spatial gradient: 24% of core-area MSAs are identified as Key-Value MSAs, while 23% of peripheral MSAs are categorized as General-Retention MSAs. Based on the classification, differentiated renewal strategies are proposed: high-potential MSAs should prioritize public space enhancement and walkability improvements, whereas low-potential MSAs should focus on upgrading basic transit infrastructure. The study provides a replicable method for classifying MSAs based on spatial and economic indicators, offering new theoretical insights and practical tools to guide evidence-based urban regeneration and station–city integration in high-density metropolitan areas such as Beijing.

1. Introduction

Transit-Oriented Development (TOD) has evolved into a fundamental urban planning paradigm, coupling public transportation networks with contiguous land use to promote sustainable urbanization [1]. A broad academic consensus has validated its effectiveness in enhancing transport efficiency and reshaping urban form [2,3], while also generating multidimensional synergistic benefits: promoting transport sustainability by reducing vehicle kilometers traveled [4,5,6,7]; optimizing spatial structure by curbing urban sprawl [8,9,10]; stimulating economic vitality through property value appreciation [11,12,13]; and improving environmental quality by mitigating air pollution [7,14].

As both a key carrier of urban regeneration and a pivotal node in the city’s transport network, the metro station area (MSA) has become central to contemporary urban transformation agendas. How to scientifically plan and regenerate these areas to optimize urban function and enhance residents’ quality of life has emerged as a critical issue and research hotspot. A core challenge in studying regeneration strategies for metro station areas (MSAs) lies in the development of precise and effective plans tailored to the highly heterogeneous characteristics of these spaces. Accordingly, identifying MSA typologies and formulating differentiated strategy frameworks has become a crucial pathway to addressing this complexity. Previous research on TOD typologies has sought to evaluate existing TOD conditions or assess the TOD potential of transit station areas [15]. Calthorpe [16] pioneered the classification of TOD into two types: “urban” communities and “neighborhood” communities. Dittmar and Poticha [17] further developed a sixfold TOD typology based on roles and functions within the station environment: urban downtown, urban neighborhood, suburban town center, suburban neighborhood, neighborhood transit zone, and commuter town. However, to design more targeted policies, planners and policymakers increasingly rely on rigorous and objective empirical assessments of existing station areas.

Some scholars have proposed classification methods based on specific built environment features of MSAs [18,19], largely grounded in the six “D” variables of TOD environments: density, diversity, design, destination accessibility, distance to transit, and demand management for car traffic [20]. Kamruzzaman et al. [21] used six TOD indicators and clustering analysis to identify four TOD types in Brisbane: residential TOD, activity center TOD, potential TOD, and non-TOD. Higgins and Kanaroglou [18] applied latent class analysis to categorize 372 MSAs in Toronto into 10 distinct neighborhood types based on 5D indicators. Other studies have classified MSAs using environmental or functional variables; for example, by major land-use function into center, district, and corridor types [22], or into employment, commercial, and residential categories [23]. Some recent work focused on classification by development potential to guide investment decisions [24]. Among the most widely adopted frameworks is Bertolini’s “node–place” model [25], which categorizes stations into five types—balanced, stressed, dependent, unbalanced-node, and unbalanced-place—based on the relative values of node accessibility and place quality. This model has become a foundational tool in evaluating MSAs and typologies. Subsequent enhancements to the model have integrated spatial syntax indices such as Closeness Centrality and Betweenness Centrality to analyze the relationship between transit conditions and land-use development [26]. However, these classifications remain incomplete and lack a universal and objective standard, often resulting in inconsistent typologies across different studies and cities, thereby limiting comparative analyses [27].

Although the existing literature has significantly advanced MSA regeneration research and provides useful references, it tends to overlook the varying impacts of transit systems on urban renewal across different city contexts. Moreover, there remains a lack of data-driven, analytical classification methods for MSAs. Current typologies primarily rely on station location, function, and transport characteristics; yet, due to significant intercity variation, such traditional methods are not universally applicable. For instance, in Beijing’s inner core (within the Second Ring Road), stringent land-use regulations constrain development intensity, leading to misalignments between a station’s location and its functional attributes. Meanwhile, Beijing’s metro network, which follows a “ring + radial” layout, is evenly distributed and has already fostered high-density residential and employment clusters on the periphery of its central city. These unique features of Beijing’s MSAs increase the complexity of classification, making conventional methods insufficient for comprehensive analysis.

This study introduces the RFM model, widely used in marketing to evaluate and classify customer value, for the classification of MSAs for the first time. Based on this model, we analyze three key indicators for 118 MSAs in Beijing: POI distribution, population density, and land price. The analysis produces eight types of MSAs: Key-Value, Key-Development, Key-Maintenance, Key-Retention, General-Value, General-Development, General-Maintenance, and General-Retention. Building on this classification, we further examine the spatial characteristics and existing challenges of each MSA type and propose corresponding renewal strategies. This classification-oriented framework provides a new theoretical lens for guiding MSA regeneration and offers methodological innovations to support evidence-based decision-making and design in urban renewal.

2. Methodology

2.1. Site Selection and Data Preparation

This study aims to explore the spatial regeneration value and potential of various MSAs. Accordingly, the built environment within the study scope was required to possess a certain degree of maturity. Under the “incremental reduction” policy of the Beijing Urban Master Plan (2016–2035), the city’s central area, located within Beijing’s Fourth Ring Road, has developed a relatively stable spatial structure and accumulated abundant stock space resources [28]. The built environments surrounding metro stations in the area have an average development history of over 15 years and display marked intergenerational discrepancies between existing functional facilities and contemporary urban demands, making spatial regeneration an urgent necessity [29]. In the selection of study sites within the central urban area, both the characteristics of the metro network and its coupling with urban spatial structure were considered. The sampling strategy included the following:

• Priority coverage of Lines 2 and 10, which run along the Second and Third Ring Roads, respectively. These ring lines connect Beijing’s core functional districts, and their surrounding areas are typically characterized by high-intensity, mixed-use development.
• Supplementary coverage of Lines 6 (east–west axis) and 5 (north–south axis), which traverse transitional zones between traditional residential areas and emerging commercial zones, thereby capturing the gradient evolution of urban spatial patterns.
• Specifically, peripheral high-maturity stations, such as Xi’erqi Station on the Changping Line and Xihong Men Station on the Daxing Line, are included. Although located outside the Fourth Ring Road, these stations exhibit a stable employment–residential balance and exceed 100,000 daily passengers, thus holding significant research value.

As of January 2025, the Beijing Metro system consists of 29 operational lines. Based on the maturity of the built environment and the urgency of regeneration, 118 representative stations were selected as study cases. Foundational spatial datasets from 10 March to 27 March 2025, including POI data, population density, and land price, were collected via multiple platforms, such as OpenStreetMap, Amap Open Platform, Figshare, and official documents from the Beijing Municipal Government. These data were subsequently integrated and processed using the QGIS platform (

Table 1. Data sources.

Data	Source	Data Year(s)	Collection Date
10 min walkable range	Open Route Service API https://www.openstreetmap.org (accessed on 10 March 2025)	2024	10 March 2025
POI data	Amap Open Platform https://lbs.amap.com (accessed on 15 March 2025)	2020, 2021, 2022, 2023	15 March 2025
Population density	Figshare Platform https://figshare.com (accessed on 20 March 2025)	2024	20 March 2025
Land price	Beijing Municipal Government https://www.beijing.gov.cn (accessed on 27 March 2025)	2022	27 March 2025

).

2.2. Definition of Station Area Scope

The scope of the MSA refers to the station’s walkable catchment, service area, or influence zone. At present, there is no universally accepted standard for delineating station areas. In this study, the MSA is defined as the zone accessible within a 10 min walking distance, based on typical walking speed and urban traffic conditions [30]. To delineate this boundary, isochrone zones were calculated using real-time travel time data obtained through the Open Route Service API. Developed and maintained by the Heidelberg Institute for Geoinformation Technology, Open Route Service provides routing services based on crowdsourced open-source data from OpenStreetMap. This tool enables researchers to generate isochrones that define the areas and buildings reachable within a given time or distance constraint.

In this study, a 10 min walking duration was set as the input, with all publicly accessible station exits serving as origin points. Isochrone polygons were generated in batches via the API. Using QGIS spatial analysis tools, the isochrones for each station’s exits were geometrically merged to produce composite polygons, which represent the final spatial extent of each station area (Figure 1). The distribution of all 118 MSAs in Beijing is shown in Figure 2.

2.3. RFM Model

The MSA, as a multifunctional urban node, serves a range of user groups in ways that are conceptually similar to customer segments in for-profit enterprises—the original application domain of the RFM model. This conceptual alignment provides theoretical support for adopting the RFM model in studies of MSA regeneration. Traditionally, the RFM model classifies customers based on three indicators: recency (R)—the time since their last purchase; frequency (F)—the frequency of purchases; and monetary (M)—the total amount spent. By comparing individual scores for each dimension to the sample mean, customers are categorized into high or low groups, generating eight distinct customer types. This classification informs strategies aimed at increasing customer retention and optimizing market targeting and resource allocation (Figure 3).

Conceptual mapping between key MSA indicators and the RFM dimensions reveals strong parallels. For instance, the recency of POI updates in an MSA reflects urban vibrancy and attractiveness, aligning with the R indicator. The frequency of pedestrian activity relates to spatial usage intensity, corresponding to the F indicator. Finally, land price—an established proxy for economic potential—mirrors the M indicator, capturing investment appeal and value appreciation potential (Figure 4).

Accordingly, this study proposes an adapted RFM framework tailored for urban spatial analysis. The revised model enables quantitative evaluation and classification of MSAs by defining three dimensions: recency (R)—represents spatial vibrancy based on functional richness; frequency (F)—captures persistent urban vitality through population clustering; and monetary (M)—measures the economic value of the station area. Each indicator is classified into high or low based on whether the normalized score exceeds the average for that dimension. Following the traditional RFM classification logic, this yields eight MSA types, each with distinct characteristics and regeneration needs (

Table 2. Metro station area (MSA) type based on the RFM model.

No.	R Value	F Value	M Value	MSAs Type
I	High	High	High	Key-Value MSAs
II	High	Low	High	Key-Development MSAs
III	Low	High	High	Key-Maintenance MSAs
IV	Low	Low	High	Key-Retention MSAs
V	High	High	Low	General-Value MSAs
VI	High	Low	Low	General-Development MSAs
VII	Low	High	Low	General-Maintenance MSAs
VIII	Low	Low	Low	General-Retention MSAs

).

2.4. Measurement of Recency (R) Based on POI Data

The recency (R) indicator is defined as the average number of POIs within the station area over the past four years. POI data—containing geographic points with attributes such as name, category, and location—offer a quantitative measure of functional diversity and urban vitality within the MSA [31]. In this study, POI data were retrieved from the Amap Open Platform for the years 2020 to 2023, covering 14 categories, including food services, retail, tourist attractions, enterprises, education and research, residential services, healthcare, and others.

After cleaning the dataset to remove invalid coordinates and duplicate entries, the valid POI counts were 237,005 (2020), 246,480 (2021), 239,603 (2022), and 221,567 (2023). The R value for each station area was then calculated as the average POI count across the four years, as shown below:

$$R={\displaystyle \frac{1}{4}}\sum _{t=2020}^{2023}{POI}_{i,t}$$

(1)

where POI_i,t is the number of POIs in station area i in year t.

2.5. Measurement of Frequency (F) Based on Population Density

The frequency (F) indicator is represented by population density within each station area, reflecting how densely populated and frequently utilized the space is [32]. The population data were sourced from the China 7th Census 100 m raster dataset published on the Figshare platform [33].

Using QGIS (version 3.34.0 ‘Maidenhead’) software, the raster data were geo-referenced and extracted based on the station area boundaries. The F value was then computed as the total population within a station area divided by its spatial area:

$$F={\displaystyle \frac{P_i}{A_i}}$$

(2)

where P_i is the total population in station area i, and A_i is its spatial area.

2.6. Measurement of Monetary Value (M) Based on Land Price

The monetary (M) indicator is defined by the land price level of the station area. Land price, as a direct reflection of market valuation, serves as a reliable proxy for economic potential and development intensity [34]. Land price data were obtained from the 2022 Beijing Benchmark Land Price Update published by the municipal government.

These data were processed in QGIS by overlaying benchmark land price zones with MSA boundaries. Each station area’s land price level was then calculated as a weighted average based on the proportion of different land price zones within its boundary. Since higher land value corresponds to lower numerical land price levels in the original dataset, the values were standardized using the formula:

$$L_j^{\prime }={\displaystyle \frac{L_{max}-L_j}{L_{max}-L_{min}}}$$

(3)

where $L_j^{\prime }$ is the normalized land price index for class j, and L_max and L_min are the maximum and minimum land price levels, respectively.

The final M value for each station area was then computed as:

$$M=\sum _j{(L}_j^{\prime }·{\displaystyle \frac{A_{ij}}{A_i}})$$

(4)

where A_ij is the area of land price level j within station area i, and A_i is the total area of station i.

2.7. Normalization of Indicators

To eliminate unit inconsistencies and improve model accuracy and comparability, all indicators were normalized. This procedure transforms variables of different scales into dimensionless values within the [0, 1] range:

$$X_i^{\prime }={\displaystyle \frac{X_i-X_{min}}{X_{max}-X_{min}}}$$

(5)

where X_i is the raw value of the indicator, and X_min and X_max represent the minimum and maximum values, respectively.

2.8. Indicator Scoring and Classification

Traditional RFM classification often uses binary grouping based on the mean or median, which may oversimplify complex distributions. To improve granularity and reduce classification bias from data skewness, this study adopts a five-quantile scoring method. After normalization, each indicator is divided into five quantiles, as follows:

• 1st quantile (0–20%): Score = 1
• 2nd quantile (20–40%): Score = 2
• 3rd quantile (40–60%): Score = 3
• 4th quantile (60–80%): Score = 4
• 5th quantile (80–100%): Score = 5

The average score of each indicator determines its classification: scores greater than or equal to the mean are classified as high, and those below as low. This binarized result is then used to assign each station area to one of the eight RFM types defined earlier.

3. Results

3.1. Results of the Recency (R) Indicator

Analysis of POI data across 118 MSAs in Beijing reveals a pronounced concentric gradient of urban function availability, characterized by high density within the Second and Third Ring Roads and a marked decline towards the periphery (Figure 5). Stations in core commercial zones exhibit a high concentration of POIs, indicating strong functional diversity, active commercial environments, and well-developed service facilities. For instance, Chongwenmen Station recorded the highest four-year average POI count of 6244, followed by other central hubs like Jintai Xizhao (7339), Guomao (5256), and Dongdan (3840). The areas surrounding these stations are characterized by a high-intensity mix of commercial, office, cultural, entertainment, and residential amenities, constituting the city’s core vibrant areas. These features attract substantial foot traffic and promote agglomeration. MSAs near the Fourth Ring Road show moderate POI density with mixed functions, whereas those beyond the Fifth Ring Road—such as Tiantongyuan Nan (S) Station and Tiantongyuan Station, with average POI counts of only 1237 and 841, respectively—are dominated by residential and office uses with limited commercial offerings. Typically located on the urban fringe, these MSAs suffer from underdeveloped infrastructure and inadequate commercial services, resulting in limited local employment and leisure opportunities. This functional scarcity results in low local vitality and high dependence on central districts for employment and leisure, forcing residents to undertake long-distance commutes.

The spatial disparity captured by the recency (R) indicator highlights not only imbalanced service distribution but also supports the classification of MSAs based on functional completeness. High R values are associated with complex and frequently updated urban environments, typifying Key-Value, Key-Development, General-Value, and General-Development MSAs. In contrast, low R values correspond to stations with poor functional diversity, which are predominantly classified as Key-Retention or General-Retention categories, indicating constrained regeneration potential.

3.2. Results of the Frequency (F) Indicator

Analysis of population density across 118 MSAs reveals a polycentric distribution pattern in Beijing, yet also highlights a critical divergence between residential concentration and sustained activity (Figure 6). Notably, MSAs encompassing large-scale residential communities, such as Tiantongyuan Nan (S) (387 persons/ha2) and Tiantongyuan (378 persons/ha2), exhibit the highest population densities, a direct manifestation of concentrated housing development. Conversely, core urban MSAs like Yonghegong (192 persons/ha2) and Dongdan (117 persons/ha2) show significantly lower resident densities due to building height restrictions, heritage preservation policies, and population decentralization efforts.

The frequency (F) indicator captures this distinction by measuring not only population aggregation but, more importantly, the permanence and intensity of human activity. A high F value signifies persistent spatial usage and is characteristic of stations where high residential density translates into consistent local vitality. Conversely, low F values—frequent in the urban core—indicate stations whose daytime influx of people is not sustained by a comparable nighttime population, reflecting a reliance on non-local visitors. Integrating the F indicator with the recency dimension enables refined identification of MSAs where high population density coexists with functional deficiency, revealing significant latent potential for targeted regeneration.

3.3. Results of the Monetary (M) Indicator

Analysis of land price levels across 118 MSAs reveals a distinct dual spatial gradient, characterized by a radial decline from the urban center outward and a pronounced north-to-south disparity (Figure 7). Land values peak in central urban districts—such as the historic core and primary business centers—with stations like Dongdan (1.50), Dengshikou (1.53), and Dongsi (1.51) exhibiting the highest levels, driven by their status as traditional commercial and cultural hubs, rich historical heritage, and superior amenities that intensify demand for properties. A clear decay in value is observed with increasing distance from the core. Areas between the Third and Fourth Ring Roads, such as Wangjing (3.56), command moderately lower prices, reflecting their transitional status with developed yet less concentrated commercial functions and slightly weaker connectivity. The gradient culminates in the peripheral zones beyond the Fifth Ring Road, including Tiantongyuan (6.09), where land prices are the lowest, characterized by a predominance of residential uses, abundant land supply, limited local employment, and weaker transport links.

The spatial differentiation captured by the monetary (M) indicator provides a critical economic basis for MSA classification. High M values are predominantly associated with the “Key” category, including high-value, centrally located stations where elevated land prices signal strong investment appeal and redevelopment potential. Conversely, low M values correspond to the “General” group, encompassing stations in peripheral or southern locations where lower economic intensity and market valuation constrain regeneration opportunities. When integrated with the recency (R) and frequency (F) dimensions, the M indicator completes a robust tripartite framework that effectively delineates the regeneration potential and strategic positioning of all eight MSA types within Beijing’s spatial system.

3.4. Classification Results of RFM Indicators

The mean and quantile scores of the RFM indicators are summarized in

Table 3. Quantiles of the three RFM indicators.

Indicator	Mean	20% Quantile	40% Quantile	60% Quantile	80% Quantile
Recency	0.250	0.109	0.173	0.247	0.400
Frequency	0.450	0.283	0.428	0.540	0.620
Monetary	0.524	0.430	0.525	0.672	0.810

. Each indicator was assigned a value category (high or low) based on whether its average score exceeded the mean. The scoring range and corresponding classification intervals are shown in

Table 4. Scoring range and value segmentation intervals.

Indicator	Score 1	Score 2	Score 3	Score 4	Score 5	Mean Score	Low Value Type	High Value Type
Recency	[0, 0.109)	[0.109, 0.173)	[0.173, 0.247)	[0.247, 0.400)	[0.400, 1]	2.992	[1, 2.992)	[2.992, 5]
Frequency	[0, 0.283)	[0.283, 0.428)	[0.428, 0.540)	[0.540, 0.620)	[0.620, 1]	3.008	[1, 3.008)	[3.008, 5]
Monetary	[0, 0.430)	[0.430, 0.525)	[0.525, 0.672)	[0.672, 0.810)	[0.810, 1]	2.898	[1, 2.898)	[2.898, 5]

.

3.5. Classification Results and Renewal Strategies

The classification results of the 118 MSAs within the final study scope are presented in

Table 5. Results of station domain classification.

Type	MSA Classification	Stations	Count
I	Key-Value MSAs	Baishiqiaonan, Chaoyang Men, Chegongzhuang, Chegongzhuang Xi (W), Daguan Ying, Dongsi Shitiao, Dongzhi Men, Fucheng Men, Haidian Huangzhuang, Huayuan Qiao, Jintai Lu, Jingsong, Jiulongshan, Liufang, Panjiayuan, Shuangjing, Suzhou Jie, Weigongcun, Xizhimen, Zhichunli, Zhichun Lu, Chongwen Men, Ciqu Kou, Hepingli beijie, Puhuangyu, Heping Xiqiao, Huixin Xijie Beikou, Huixin Xijie Nankou	28
II	Key-Development MSAs	Anding Men, Anzhen Men, Beihaibei, Beijing Zhan (Beijing Railway Station), Dongdaqiao, Fuxingmen, Guomao, Heping Men, Hujialou, Jishuitan, Jianguomen, Jiandemen, Jintai Xizhao, Liangma Qiao, Nanluogu Xiang, Nongye Zhanlanguan (Agricultural Exhibition Center), Ping’anli, Qianmen, Tuanjiehu, Xuanwu Men, Changchun Jie, Beixin Qiao, Dongdan, Dongsi, Dengshi Kou, Yonghegong (Lama Temple), Zhangzizhong Lu, Temple of Heaven East Gate	28
III	Key-Maintenance MSAs	Fangzhuang, Muxidi, Taoranting	3
IV	Key-Retention MSAs	Aolinpike Gongyuan (Olympic Park), Beitucheng, Gongzhufen, Gulou Dajie, Lianhua Qiao, Sanyuan Qiao, Shaoyaoju, Taiyanggong, Tianqiao, Xidiaoyutai	10
V	General-Value MSAs	Jiaomendong, Mudanyuan, Qingnian Lu, Shiliuzhuang, Wangjing, Wudao Kou, Xitucheng, Liujiayao, Songjiazhuang	9
VI	General-Development MSAs	Fengtai Science Park, Jiangtai, Shilihe, Changchun Qiao, Beiyuanlu Bei (N)	5
VII	General-Maintenance MSAs	Haidian Wuluju, Jiaomen Xi (W), Shilipu, Capital Univ. of Economics and Business, Wukesong, Lishui Qiao, Tiantongyuan Nan (S), Tiantongyuan	8
VIII	General-Retention MSAs	Bagou, Caoqiao, Chedaogou, Chengshou Si, Cishou Si, Dahong Men, Fenzhongsi, Fengtai Railway Station, Happy Valley, Huoqiying, Huoying, Jijiamiao, Lincuiqiao, Liuli Qiao, Niwa, Rongjing Dongjie, Shangdi, Xi’erqi, Xihong Men, Xiju, Xixiaokou, Xingong, Yizhuang Wenhuayuan (Yizhuang Culture Park), Yuanmingyuan Park, Lishuiqiao Nan (S), Datunludong, Tiantongyuan Bei (N)	27

, and their spatial distribution is illustrated in Figure 8.

Based on the three-dimensional RFM framework, the classification reveals a concentric pattern of decreasing station value from center to periphery, reflecting a structure of “dense urban core + ring-based transport hubs.” Among the 28 Key-Value MSAs, 14 are transfer stations. Their average R, F, and M scores are 0.41, 0.60, and 0.76, significantly exceeding the overall averages (0.25, 0.45, 0.52). However, stations like Weigongcun show high M (0.81) but low R (0.21), indicating insufficient commercial activation despite high traffic, suggesting a need for value transformation strategies.

The Key-Development and Key-Maintenance MSAs (28 and 3 stations, respectively) are mostly located in historic urban zones and show lower rates of transfer connections, reflecting preservation-oriented development. For instance, Fangzhuang Station (Key-Maintenance) shows high F (0.59) and M (0.62) but low R (0.16), highlighting issues such as weak on-site consumption and declining functionality, pointing to a lack of commercial infrastructure. Key-Retention MSAs (10 stations), such as Gulou Dajie, display high M (0.81) but extremely low R (0.13), revealing that heritage protections restrict commercial development, limiting cultural–commercial synergy. Among General-Value MSAs, stations like Qingnianlu show relatively high F (0.68) but low M (0.45), indicating lagging commercial performance in large-scale residential zones with job–housing separation. General-Development MSAs, such as Shilihe Station, exhibit relatively high R due to niche market clusters (e.g., antique and plant–pet markets). General-Maintenance MSAs like Tiantongyuan Nan (S) show the highest F (1.0) but the lowest M (0.05), as only 12% of 250,000 daily commuters consume locally, exposing severe value leakage caused by spatial–functional mismatch. The most vulnerable category, General-Retention MSAs (25 stations), shows the lowest average scores across all dimensions (R = 0.08, F = 0.25, M = 0.29). These include Fenzhongsi and Tiantongyuan Bei (N) Station, forming a contiguous low-value belt on the urban fringe that requires urgent intervention to prevent functional collapse.

Renewal strategies based on station classification include the following:

• Strategies for high-potential MSAs. Key-Value MSAs, possessing the highest composite value and renewal potential, should be prioritized in regeneration plans. These areas warrant multi-dimensional upgrades to ensure efficient and sustainable revitalization. Renewal should emphasize the following:
  • Optimization of public spaces;
  • Enhancement of slow-mobility systems;
  • Creation of distinctive urban landscapes.

• The goal is to transform transit-centric spaces into multifunctional environments integrating social, cultural, and service-oriented uses. These areas should also promote green mobility and celebrate local identity, becoming catalytic cores that drive broader urban regeneration.

2.

Strategies for moderate-potential areas. The remaining six types—Key-Development, Key-Maintenance, Key-Retention, General-Value, General-Development, and General-Maintenance MSAs—possess moderate renewal potential. Interventions should focus on improving underperforming indicators, as follows:

• For low R scores, introduce new functions and activities to stimulate vibrancy;
• For low F scores, enhance accessibility and attractiveness of public spaces to increase foot traffic;
• For low M scores, upgrade infrastructure and reconfigure land use to boost economic appeal.

3.

Strategies for low-potential Areas. General-Retention MSAs, with minimal renewal potential, should focus on reinforcing their basic function as transportation nodes. Priority should be given to improving road infrastructure and enhancing transit connectivity. These interventions aim to prevent spatial decline and support the stability of urban peripheries.

3.6. Detailed Strategies for Each MSA Type

For precise planning and decision-making,

Table 6. A practical guide to renewal strategies for different MSA types.

MSA Type	RFM Profile and Core Challenges	Primary Planning Objective	Targeted Strategies and Key Actions
Key-Value MSAs	High R, High F, High M Dense and diverse POIs, high population density, and premium land value, yet challenged by fragmented public spaces and inefficient feeder systems.	Quality Enhancement and Value Sustainment: To transform into exemplary, sustainable, and multifunctional urban cores.	Optimize Function (R): Integrate underground spaces to connect key POIs; introduce high-value-added functions like curated retail and art galleries. Manage Flow (F): Create pedestrian-priority zones and optimize bus-feeder services to disperse peak-hour crowds. Sustain Value (M): Establish urban design review boards to enforce esthetic codes and ensure long-term economic resilience.
Key-Development MSAs	High R, Low F, High M Rich POIs and high land value, but low resident density leads to unsustainable tourist-dominated vibrancy.	Vibrancy Cultivation with Sensitivity: To stimulate round-the-clock activity while preserving heritage integrity.	Increase Residency (F): Facilitate adaptive reuse of historic structures into boutique hotels or apartments to boost nighttime population. Improve Experience (R): Implement heritage trails and smart wayfinding systems to manage tourist flow and enhance POI accessibility. Diversify Activity: Support local creative industries and seasonal markets to attract daily visits from residents.
Key-Maintenance MSAs	Low R, High F, High M High population density with outdated and monotonous POIs leads to reliance on external centers for services.	Functional Completeness and Livability: To develop self-sufficient 15 min neighborhoods.	Add Functions (R): Encourage commercial conversion of ground floors in residential blocks for daily needs (e.g., grocers, clinics, cafés). Enhance Amenities: Transform underused spaces into community gardens and sports facilities to elevate public space quality. Foster Balance: Introduce small-scale co-working spaces to promote local job–housing balance.
Key-Retention MSAs	Low R, Low F, High M Cultural/ecological constraints limit POI diversity and population density, while incurring heavy traffic pressure.	Efficiency and Heritage Symbiosis: To achieve precise traffic management and context-sensitive activation.	Alleviate Congestion: Deploy intelligent traffic signals and dedicated tourist shuttle routes to minimize local disruption. Cultural Activation (R): Introduce low-impact cultural POIs (e.g., artisan workshops, tea houses) within protected buffers. Temporal Programming: Develop night exhibitions or light shows to boost off-peak visitation (F).
General-Value MSAs	High R, High F, Low M Strong POI and population base, but lacks high-end functions, resulting in underperforming land value.	Value Leap and Sub-Center Creation: To become attractive regional commercial destinations.	Upgrade Function (R): Facilitate land reassembly for mixed-use complexes with flagship retail and Grade-A offices. Create Identity: Develop themed commercial streets and attract flagship stores to build district branding. Monetize Potential (M): Leverage high footfall to promote night-time economy, converting into economic gains.
General-Development MSAs	High R, Low F, Low M Emerging specialized POI clusters exist, but low resident density stifles overall economic value.	Investment Catalysis and Nurturing: To accelerate maturation and attract residents and capital.	Boost Density (F): Prioritize residential land supply with moderate density bonuses to attract new populations. Protect Character (R): Enhance existing specialized POI clusters (e.g., antique markets), avoiding homogenization. Build Foundations: Invest first in high-quality schools and clinics to lay the groundwork for future value (M) appreciation.
General-Maintenance MSAs	Low R, High F, Low M Dormitory towns with high population density but severe lack of local jobs and services, causing massive commutes.	Burden Reduction and Optimization: To enhance self-sufficiency and optimize commuting efficiency.	Generate Local Jobs: Zone for clean office/R&D parks to create local employment opportunities. Add Services (R): Activate marginal spaces for essential commerce (convenience stores, markets) and services (gyms, tutoring). Optimize Transit (F): Launch express shuttle buses to major job centers and refine microcirculation bus routes.
General-Retention MSAs	Low R, Low F, Low M Low scores across all dimensions, with minimal POIs, low density, and the lowest land value.	Stabilization and Cost-Effective Improvement: To ensure reliable transit service and basic safety with minimal investment.	Ensure Core Function: Maintain infrastructure for safety, cleanliness, and accessibility; improve transfers to regional hubs. Low-Cost Upgrades: Implement cost-effective improvements like better lighting, surveillance, and weather protection. Preserve Flexibility: Control costs and avoid over-investment, reserving land for potential long-term transformation.

provides a clear and concise summary of strategic priorities and key actions for each of the eight MSA types identified in this study.

4. Discussion

This study developed a station-area classification framework based on the RFM model, incorporating three core indicators—recency (R), frequency (F), and monetary value (M)—corresponding to POI count, population density, and land price level, respectively. By processing and analyzing multi-source spatial data using QGIS, 118 MSAs in Beijing were systematically categorized into eight distinct types. The characteristics of each category were analyzed in depth, leading to the formulation of tailored renewal strategies.

In the context of urban regeneration, it is insufficient to merely distinguish between MSAs and non-metro areas. Rather, it is more productive to examine the differences among various types of MSAs. For instance, Key-Value MSAs, characterized by high development intensity, superior accessibility, and rich functional mix, demonstrate clear advantages in centrally located commercial hubs. These areas concentrate working and residential populations, with strong demand for both housing and commercial services. The multi-dimensional advantages of Key-Value MSAs enable them to effectively meet these needs, making them highly adaptive and responsive to complex urban demands. Building on this insight, we propose a mechanism-based upgrading framework for other MSA types, ensuring they can replicate such commercial vitality. The framework incorporates four key levers, as follows:

• Urban morphology optimization: For classes with low spatial permeability, encourage block-scale land readjustment to increase plot diversity and enable mixed-use, TOD-compatible development (policy tools: FAR bonuses, land transfer premium reductions).
• Activity programming: For culturally significant station areas, introduce branding strategies that integrate cultural identity with commercial functions—e.g., leveraging intangible heritage (Dongsi calligraphy culture) to create periodic cultural markets.
• Accessibility-oriented redevelopment: For node-dominant types, implement pedestrian-priority zones and station-anchored street commerce to replicate the footfall patterns of high-performing stations.
• Public–private partnership (PPP) models: For service-deficient classes, use PPP schemes to attract private investment for retail and community services within 500 m catchment zones, reducing sole reliance on public funds.

Together, these interventions translate the strengths of MSAs into actionable strategies for underperforming station areas, guiding them toward comparable economic and social success.

This finding is consistent with the work of Dan et al. [35], who analyzed 347 metro stations in Shanghai and identified five station clusters based on strong correlations between passenger volume and three variables: transportation (T), pedestrian-oriented accessibility (O), and urban development (D). Stations within the high-TOD-performance cluster scored highly across all three dimensions and were predominantly located in historic urban cores. This pattern aligns with the characteristics of Key-Value MSAs identified in the present study, confirming their central location and robust performance across domains.

Similarly, Su et al. [3], using the Node–Place–Function model in combination with interpretable machine learning, revealed the spatiotemporal heterogeneity of functional mixing. Their findings underscore the importance of urban functionality, offering empirical support for the classification of Key-Maintenance MSAs. These station areas, typically located in mature residential zones, tend to have low functional diversity and are dominated by residential uses, limiting the development of supporting commercial services. Renewal strategies for such areas should prioritize meeting daily living needs, optimizing transportation networks, and enhancing public service provision to improve residents’ quality of life.

Further comparative insights can be drawn from Lyu et al. [36], who categorized Beijing metro stations into six types based on three TOD dimensions. One category, corresponding to the General-Development MSAs in this study, includes stations located on the periphery of the urban core with moderately low TOD scores but a high degree of functional mix. For these stations, renewal strategies should emphasize increasing urban density and improving walkability to support the already high level of functional integration—echoing the findings of the present study.

Nevertheless, gaps remain in the existing literature. For example, Liu et al. [37] used machine learning to optimize TOD classification in Ningbo but did not account for station areas within historic preservation zones, thus overlooking special classifications such as heritage-sensitive station areas. Similarly, Papa et al. [38] proposed a simple and replicable GIS-based method for TOD classification and identified five types: undeveloped stations, employment-oriented stations, mixed-use central stations, medium-value residential stations, and high-value residential stations. These align well with several types defined in this study—such as General-Retention MSAs, General-Value MSAs, and Key-Development MSAs—but do not address specific station types like those with cultural heritage protection or transport-hub overlays.

The findings of this study also provide valuable insights for the planning of new metro stations. Specifically, we extend the discussion to address how new stations can facilitate urban development, how to ensure that anticipated residential and commercial benefits outweigh the associated costs, and what criteria should be prioritized when selecting locations for new station deployment.

New metro stations contribute to urban development by activating revitalization in service-deficient corridors, reducing integrated travel costs to employment centers and public services, and fostering the formation of compact, mixed-use TODs around station areas. To ensure that economic and social benefits outweigh costs, it is essential to tie project approval to verifiable evidence derived from passenger flow projections and anticipated land value appreciation. During the pre-construction evaluation phase, monetized benefits—including fare revenue, travel time savings, and emission reductions—should be integrated with land value capture mechanisms. This approach allows a portion of the projected value increment to be allocated toward covering capital expenditures, thereby ensuring that each new station achieves both financial sustainability and significant social benefits. Accordingly, site selection for new stations should prioritize areas with inadequate public service coverage yet high redevelopment potential, thereby activating underperforming urban corridors and promoting job–housing balance [39]. An evaluation model incorporating multi-source spatiotemporal data and land value forecasting enables refined estimation of passenger flow benefits and renewal potential [3]. Ultimately, through land value capture instruments, anticipated appreciation revenues can be harnessed to offset a share of construction costs, ensuring that projects deliver transport efficiency and long-term financial viability throughout their lifecycle [40].

Although this research contributes to the literature on MSA classification, several limitations should be acknowledged. The scope of the study was geographically constrained to selected areas in Beijing. While diverse urban zones were included, the sample size remains limited and should be expanded in future work. Nevertheless, the methodology developed in this study demonstrates potential applicability to other cities with similar high-density urban contexts, provided that comparable data sources are available. In addition, many of the data were sourced online. Although timely and comprehensive, the lack of on-site validation may affect the explanatory depth of the indicators. Future research could explore the approach’s adaptability under different data conditions, including integrating traditional survey methods with emerging digital data sources for enhanced robustness.

Moreover, the renewal strategies proposed in this study, while systematically derived, would benefit from further refinement to enhance their specificity and feasibility. In particular, it should be noted that passenger flow dynamics at different times of day and across weekdays versus weekends/holidays are critical to capturing the commercial behavior of metro users and their implications for planning strategies. Due to data availability constraints, such fine-grained passenger flow indicators were not incorporated into the present analysis. However, future research will seek to refine the granularity of data by integrating daily and hourly passenger flow statistics, thereby enabling a more nuanced assessment of station functions and commercial vitality.

In addition, future research will aim to expand the coverage to MSAs across multiple metro lines and integrate additional indicators for a more comprehensive evaluation. The proposed framework also offers potential for supporting cross-city comparisons and dynamic monitoring of urban development patterns over time, which could significantly enhance policy-relevant insights. Particular attention will be paid to strengthening the internal logic between classification outcomes and renewal strategies, ensuring that proposed interventions are both evidence-based and actionable. Through such efforts, this line of inquiry will provide a more precise and comprehensive foundation for guiding the renewal and development of MSAs in high-density urban contexts.

5. Conclusions

Against the backdrop of “stock-oriented development” and “station–city integration,” this study innovatively applied the RFM model to construct a systematic framework for the classification and renewal of MSAs. The proposed framework offers effective support for accurately identifying renewal potential and promoting integrated station–city development scientifically. The key conclusions are as follows:

• This study introduces the RFM model from marketing theory into the field of MSA classification, establishing a multidimensional identification framework based on vitality (R), agglomeration (F), and value (M). Empirical analysis of 118 stations in Beijing validated the model’s applicability and effectiveness, offering a novel theoretical tool and methodological pathway for assessing renewal potential and guiding classification.
• Using the improved RFM model, 118 MSAs in Beijing were categorized into eight types based on renewal potential. The results reveal a distinct spatial ring structure. The central urban core is dominated by Key-Value MSAs (24%, e.g., Chongwen Men Station) and Key-Development MSAs (24%, e.g., Dongsi Station); residential areas between the Second and Third Ring Roads feature Key-Maintenance MSAs (2%, e.g., Fangzhuang Station); Key-Retention MSAs between the Third and Fourth Ring Roads are characterized by General-Value MSAs (8%, e.g., Wangjing Station); and outer-ring areas are primarily composed of General-Maintenance MSAs (7%, e.g., Tiantongyuan Station) and General-Retention MSAs (23%, e.g., Tiantongyuan Bei (N) Station). These classifications were derived through quantitative analysis of POI density, population density, and land price levels.
• Renewal strategies tailored to each of the eight MSA types were proposed to guide precise and context-sensitive interventions. These strategies were developed around three core dimensions—functional optimization, mobility enhancement, and spatial quality improvement—but are differentiated based on the specific conditions and potential of each category. For instance, Key-Value and Key-Development MSAs emphasize high-quality public space design and pedestrian prioritization to further elevate their already high vitality; General-Value and General-Development MSAs focus on functional restructuring and transport connectivity to stimulate economic momentum; and General-Maintenance and General-Retention MSAs require foundational upgrades in infrastructure and transit service to ensure basic functionality and discourage further decline. This refined strategy framework allows for targeted and efficient renewal actions that align with the distinct roles and opportunities of each station area within the broader urban system.