Urban railways have a profound influence on urban development, affecting various aspects of a city’s growth, infrastructure, economy, and overall quality of life. They promote multi-centered or polycentric development by enhancing land use and population density as well as accessibility to different parts of the city [5,6,7]. Urban railways also play a crucial role in alleviating traffic congestion [8] and curbing urban sprawl [9]. By providing an attractive alternative to car-based commuting, they reduce vehicle kilometers traveled and contribute to improved traffic flow, shorter commute times, and reduced pollution. However, some studies found railways having varying effects at some locations within station buffer areas [10].

One of the significant impacts of urban railways is the uplift in land value and/or property value [4,11,12,13]. Proximity to railway stations typically leads to increased property values [14,15]. This effect can significantly influence property values, rendering areas served by rail transit systems more attractive to both residents and investors [16]. The influence of rail transit on property value may vary at different stages of the project. In Hong Kong, a study reported a continuous increase in property values since the construction was announced, with values even rising further after the project began [17]. In contrast, a study in Sydney found a negative impact during the project announcement phase but observed a positive trend after construction commenced [18].

Network centrality, a concept in network theory, evaluates the importance of nodes within a network. It encompasses various measures, including degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, and PageRank. Network centrality analysis finds applications in diverse fields, such as social network analysis, information science, and transportation analysis. Highly central transportation nodes, including roads, railways, or waterways, often serve as pivotal hubs, enhancing the efficient movement of passengers and goods.

In the context of railway network centrality analysis, previous studies have employed various centrality measures, yielding implications for railway network development, operation, and management. For instance, one study assessed 28 metro systems worldwide using betweenness centrality, leading to recommendations for mitigating overcrowding [21]. Another study in Shanghai compared urban railway stations using degree, closeness, and betweenness centralities, offering operational insights [22]. In Hong Kong, the rapid transit network’s evolution was ranked based on centrality measures, guiding station management and maintenance [23]. Stockholm’s urban railway network was evaluated using degree, closeness, and betweenness centralities [24], while regional and intercity railways, including China’s high-speed rail network, were assessed and grouped by centrality measures [25].

Furthermore, certain studies have explored the association between transport network centrality and other factors. Tokyo discovered a strong relationship between railway network centrality and ridership [26]. In Beijing, bus networks exhibited a high correlation with passenger flow based on centrality measures [27]. Moreover, railway network centrality has been associated with subcenter formation in polycentric cities [28].

While there are numerous studies on the impact of road network centrality, such as [29,30,31], there are relatively few studies examining the influence of railway network centrality on property values in certain cities. For instance, in Hong Kong, the influence of closeness centrality was explored [32], and in Shanghai, the focus was on degree centrality [33]. In the Scania region, the most southern region of Sweden, a study found that the centrality of the regional train network, specifically degree and closeness centralities, influenced single-family house prices, though betweenness centrality did not show a statistically significant influence [34]. On a larger scale, the centrality of China’s high-speed rail network was also found to affect land values and housing prices [35,36].

A hedonic price model refers to an econometric model used to estimate the relationship between the price of a product or service and the various attributes or characteristics that influence that price. Hedonic price models are widely used in economics and marketing to understand how consumers make choices and how prices are determined in markets with differentiated products. They are also used for various purposes, including assessing the impact of environmental attributes on property values, predicting the price of new products, and conducting cost–benefit analyses for public policy decisions.

In the context of real estate, a hedonic price model serves as a valuable tool to assess how various factors, such as location, size, the number of bedrooms, and other features, impact the price of a house or condominium unit. By analyzing a dataset of property values along with their attributes, the model provides insights into the value that buyers place on each of these characteristics.

The hedonic price model typically treats the value of real estate property as a dependent variable, influenced by its constituent attributes or characteristics, which serve as explanatory variables. Dependent variables can take various forms, including the advertised or listed price [37,38], assessed value [39], or the actual sale transaction price [11,40,41,42,43,44]. Explanatory variables are often categorized into four main groups: structural characteristics, locational characteristics, neighborhood characteristics, and transport accessibility [45].

Structural characteristics encompass various aspects of the property, including its size, age, room types, number of bedrooms, and building-related features such as the building’s height, car parking availability, shared facilities, and more.

Locational characteristics involve factors related to the property’s proximity to urban or town centers, which may include the central business district (CBD) [14,15,46,47,48] or subcenters [43,45,49,50,51], as well as proximity or accessibility to essential services like education, healthcare, public parks, retail options, and more.

Neighborhood characteristics pertain to the area’s features in the vicinity, often including activities such as employment or retail shops, along with attributes associated with transit-oriented development (TOD) environments. These attributes may include mixed land use or mixed activities [49,52,53], land use intensity [45], job–housing balance, or the density of certain population groups [12,54].

Transport accessibility factors encompass the proximity to various transportation facilities, including rail transit stations [11,37,55,56], rail services [48], bus stops [57,58], bike-sharing stations [58], bus frequency [59], major highways, and more. The proximity of a rail transit station could be measured in various ways, such as Euclidean or straight-line distance [47,48,54,60,61], distance along transportation networks [41,49,62], or other impedance measures, like travel time [40]. Furthermore, the quality of railway services, which includes factors like train frequency, travel time between stations, and overall travel convenience, has been shown to have a substantial impact on property value [48,59,63].

Spatial effects refer to the influence of the spatial arrangement or location of data points on the dependent variable. These effects can manifest in two primary forms: spatial dependence and spatial heterogeneity [64,65]. Spatial dependence pertains to the spatial relationship between values of a variable for two locations that are some distance apart. Spatial heterogeneity, on the other hand, relates to the uneven distribution of a variable’s values across space. These spatial effects can be incorporated into the hedonic price model through various techniques, including spatial lag models, spatial error models, combined spatial lag and error models, and geographically weighted regression models.

The spatial lag model takes into consideration the spatial dependencies among observations by introducing a lagged dependent variable as an additional explanatory variable in the regression model [10,11,29,49,58,66]. This lagged variable represents the average value of the dependent variable in neighboring locations, effectively acknowledging that the value of the dependent variable in one location may be influenced by the values in nearby locations. Conversely, the spatial error model considers that there is spatial autocorrelation in the error terms of the regression model, so the spatial autocorrelation is explicitly incorporated into the error term through spatially weighted values derived from nearby observations [20,34,67]. This model acknowledges that observations in close proximity may share unobserved characteristics that impact the dependent variable. Moreover, centrality was found to be incorporated in regression in both ways: spatial lag and spatial error models, for example [31,34]. Both the spatial lag model and spatial error model yield a unified set of variable coefficients and spatial parameters across the entire study area, categorizing them as global models.

In contrast to these global models, geographically weighted regression (GWR) represents a spatial regression technique that accommodates variations in the relationship between the dependent variable and explanatory variables across different spatial locations [4,43,48,61,62,68,69]. This phenomenon, known as nonstationarity, implies that parameter estimates vary across the study area [70]. Instead of estimating a single set of coefficients for the entire dataset, GWR computes a distinct set of coefficients for each location. This approach allows for the capture of spatial heterogeneity in the relationships between variables, making GWR a family of local models. GWR has been employed as a hedonic price model to examine the influence of rail transit on property value [4,42,48,71,72].