Understanding the effect of the urban built environment on online car-hailing ridership is crucial to urban planning. However, how the effects change with the analysis scales are still noteworthy. Therefore, a multiscale exploratory study was conducted in Chengdu, China, by using the stepwise regression selection and three spatial regression models. The main findings are summarized as follows. First, as the grid size increases, the number of built environment factors that have significant effects on trip intensity decrease continuously. Second, the effects of population density and road density are always positive from the 500 m grid to the 3000 m grid. As the analysis scale increases, the effect of proximity to public transportation shifts from inhibitory to facilitation, while the positive effect of land-use mix becomes stronger. Land-use type has both positive and negative effects and shows different characteristics at different scales. Third, the effects of built environment factors on online car-hailing trip intensity show different spatial variability characteristics at different scales. The effect of population density gradually decreases from north to south. The effect of road network density shows circling and wave patterns, with the former at relatively fine scales and the latter at relatively coarse scales. The spatial variation in the effect of land-use mix can only be observed more significantly at a relatively coarse scale. The effect of bus stop density is only obvious at the relatively fine and medium scales and shows a wave-like pattern and a circle-like pattern. The effect of various land-use types shows different spatial patterns at different scales, including wave-like pattern, circle-like pattern, and multi-core-like pattern. The spatial variation in the effects of various land-use factors gradually decrease with the increase in the analysis scale.

Dockless bike-sharing systems have become one of the important transport methods for urban residents as they can effectively expand the metro’s service area. We applied the ordinary least square (OLS) model, the geographically weighted regression (GWR) model and the multiscale geographically weighted regression (MGWR) model to capture the spatial relationship between the urban built environment and the usage of bike-sharing connected to the metro. A case study in Beijing, China, was conducted. The empirical result demonstrates that the MGWR model can explain the varieties of spatial relationship more precisely than the OLS model and the GWR model. The result also shows that, among the proposed built environment factors, the integrated usage of bike-sharing and metro is mainly affected by the distance to central business district (CBD), the Hotels-Residences points of interest (POI) density, and the road density. It is noteworthy that the effect of population density on dockless bike-sharing usage is only significant at weekends. In addition, the effects of the built environment variables on dockless bike-sharing usage also vary across space. A common feature is that most of the built environment factors have a more obvious impact on the metro-oriented dockless bike-sharing usage in the eastern part of the study area. This finding can provide support for governments and urban planners to efficiently develop a bike-sharing-friendly built environment that promotes the integration of bike-sharing and metro.

The spatial decomposition of demographic data at a fine resolution is a classic and crucial problem in the field of geographical information science. The main objective of this study was to compare twelve well-known machine learning regression algorithms for the spatial decomposition of demographic data with multisource geospatial data. Grid search and cross-validation methods were used to ensure that the optimal model parameters were obtained. The results showed that all the global regression algorithms used in the study exhibited acceptable results, besides the ordinary least squares (OLS) algorithm. In addition, the regularization method and the subsetting method were both useful for alleviating overfitting in the OLS model, and the former was better than the latter. The more competitive performance of the nonlinear regression algorithms than the linear regression algorithms implies that the relationship between population density and influence factors is likely to be non-linear. Among the global regression algorithms used in the study, the best results were achieved by the k-nearest neighbors (KNN) regression algorithm. In addition, it was found that multi-sources geospatial data can improve the accuracy of spatial decomposition results significantly, and thus the proposed method in our study can be applied to the study of spatial decomposition in other areas.