Detecting “Hot Spots” of Facility POIs Based on Kernel Density Estimation and Spatial Autocorrelation Technique

The distribution pattern of urban facility POIs usually forms clusters (i.e. "hot spots") in local geographic space. The kernel density estimation (KDE), which has been usually utilized for expressing these spatial characteristics, is one of the most popular visualization tools. Considering the missing of quantitative statistical inference assessment in KDE, this paper proposes a novel method to detect the hot spots of urban facility POIs. First, this method computes the attribute value of geographic unit with the "distance decay effect", then by adopting the statistical index of Getis-Ord G_i^*, we analysis the local spatial cluster characteristics of urban facilities. Comparing this method with the conventional spatial autocorrelation based on the Quadrat clustering, the attribute value of kernel density computing can preserve the local information of data, and the spatial cluster characteristics of urban facilities can reflect the continuity characteristics of urban services, for that the KDE considers the regional impact based on the First Law of Geography. The actual data experiment for analyzing the financial POIs' distribution patterns indicates that this approach is effective to extract the hot spots of urban facility POIs in city areas.