Mining spatial association rules in spatially heterogeneous environment

Abstract

This paper proposed a novel method, scaleable moving window (SMW), to mine spatial association rules while fully taking the fact that spatial heterogeneity may widely exist. SMW is based on the traditional association rule mining algorithm, Apriori. However, we integrated different moving windows (in terms of window size and shape) with Apriori. During the spatial association mining process, various sizes of windows were used to move over the region of interest. Each window, after moving through the whole region, will produce a set of association rules within the current location of the moving window. The spatial association pattern was represented by the support value and confidence value of spatial association within all the locations of the moving window by Apriori algorithm. Different windows were tested to compare the effectiveness of the windows. Compared with traditional method where the spatial association was assumed to be for the whole region, the proposed method could well reflect the reality by giving the fact that spatial association spatially varies when spatial heterogeneity exists. This proposed method was applied in the provincial capital city, Wuhan, Hubei province, China where the spatial associations between residential buildings and roads showed spatially varied, which reflected the real condition of the city.