Spatial data mining and 3D visualizing the intangible features of urban shopping areas: results from Ximending District of Taipei City

Environmental product variety is a significant factor to generate agglomeration economies for any shopping area. However, the intangibility of these multifaceted environmental externalities sometimes limited researchers and practitioners to mathematical models or 2 dimensional visibilities, which could not give a holistic overview. This study, thus, aims to establish a methodology process to generate three-dimensional(3D) visualization of the spatial distribution of the previously intangible environmental product variety characteristics. Using geostatistical and spatial analysis tools provided by GIS, the hotpots of specific attributes such as higher rental value, product variety, and shop density, could be revealed. The algorithms of PCA and CA were used for further spatial data mining, to extract the core and periphery product variety layers and the unveiling the different features among clusters. This research compared several shopping areas from the cities of Taiwan, each with over 1200 points of interest (POIs) field surveyed data from 2014-2015. The results showed that this spatial extracting model could successfully illustrate the 34 layers of retail agglomeration areas in 3D maps. The arranged core and periphery layers clearly unveil the spatial distribution of the intangible environmental variety. However, although the regression showed that higher variety can generate higher rents, the spots with highest rents does not with the highest number of product variety layers but closely link with the core layers of product variety and the highest pedestrian spots, such as the main MRT entrance. Spatially dominant anchor stores, such as department stores and cinemas, sometimes could generate sub-centres with higher rents.