Point Pattern Analysis of ATMs Distribution Based on Ripley’s K-Function Method in Nanjing City

Wang Jiechen,Lu Min,Yuan Zhenyu,Rui Yikang,Qian Tianlu

DOI: https://doi.org/10.13249/j.cnki.sgs.2016.12.009

2016-01-01

Abstract:Since distributions of many types of urban objects are not random but in some particular patterns, an-alyzing and revealing the spatial distribution pattern of these points in urban space are essential to understand social, economic and geographical factors behind the distributions, and analysis results are conductive to wide applications such as facility layout and aided decision support. In point pattern analysis, the results may be bi-ased by merely calculating the nearest neighbor distance. The Ripley’s K-function was therefore proposed with advantages of considering the distance between any pair of points. Because many urban points associated with human activities are constrained by road networks, a network K-function, as an extension of traditional planar K-function, is then presented by applying a network distance, i.e., the shortest path distance between any pair of two points. In this article, the Ripley’s K-function was applied to analyze the spatial distribution characteristics of ATMs in Nanjing City. First of all, we used both planar univariate K-function and network univariate K-func-tion to analyze the distribution pattern of ATMs at spatial scales of downtown areas and main urban districts. Then we used planar bivariate K-function and network bivariate K-function to investigate the spatial correlation between ATMs and metro stations in main urban districts. Local bivariate (cross) K-functionwas finally applied to explore the impact of metro stations on the ATMs in local areas. The results show that ATMs are highly clus-tered in both planar and network space and the cluster characteristic is more significant in downtown areas than in main urban areas. Besides, ATMs and metro stations are highly correlated in the study area. With the increase of the measuring distance, the relationship between ATM and metro station distributions shows more obvious characteristic of spatial aggregation within a certain distance. In the analysis of local cross K-function method, ATMs are clustered with metro stations in downtown areas while there is no significant clustering characteristic between ATMs and metro stations on the outskirts. It implies that the distribution of ATMs is mainly determined by the regional commercial development. In addition, for spatial pattern analysis on point objects distributed along road networks, network K-function method is more practical than planar K-function in terms of revealing appropriate distribution pattern and relationship between two types of point objects.