A statistical method for analyzing agglomeration zones of co-location between diverse facilities on a street network

Abstract

Many geographic studies examined the spatial relationships between store locations that promote urban growth through agglomeration effects. One of the key relationships is co-location, the spatial proximity among different types of facilities within a specific distance. To analyze co-location with statistical inference, several studies have employed the cross K function method. However, the ordinary cross K function method is likely to be unsuitable because it assumes a continuous and homogeneous two-dimensional plane, while stores are usually located on a street network. Another drawback is that it does not directly specify how the distances between different types of stores influence their mutual attraction over space, referred to as the agglomeration zone in this article. To address these limitations, this article develops a new statistical method named the incremental network dual K function. The usefulness of the method is illustrated by an empirical analysis of store locations in a central area of Tokyo.