A Cluster Analysis for Housing Submarkets Considering Spatial Autocorrelation

Abstract

A housing market in an urban area is not just a single market but a combination of regionally different submarkets. This study begins with a critical mind that previous researches did not consider the spatial autocorrelation of each area where the housings are located. The clustering analysis of housing submarket which considers spatial autocorrelation is performed as it follows. First, 4 housing market attribute variables are reducted to 1 variable by principle component analysis. Then, after calculating Gi*max by AMOEBA, 7 housing submarkets which have similar characteristics based on Gi*max are classified. The characteristics of each submarket are investigated, then political implication is deduced as the following. Different level of housing policy should be made to each cluster because each cluster has different level of spatial autocorrelation.