Statistical Analysis of the Pattern Similarity between Two Sets of Regional Clusters

This paper shows a statistical method of testing the pattern similarity between two sets of regional clusters. First, to measure the similarity the distance between two sets of clusters is defined. Second, a statistical method is formulated to test whether or not the distance between two sets of clusters is close. Third, the concept of spatially well-grouped clusters is defined in conjunction with the above statistical test, and it is shown that this concept is useful for regional analysis. Last, the sets of regional clusters in the Kyushu region is studied by the above statistical test and concept.