SPATIAL INFLUENCE MODEL LINKING TO LOGIT MODEL FOR THE DISTRIBUTION OF RESIDENTIAL ACTIVITIES : Spatial influence model for categorical explaining variable

Abstract

In the previous papers we proposed the spatial correlation analysis method for analyzing two-dimensional urban lattice data, and reconstructed its theory for categorical variables. This paper describes a method for estimating the spatial influence model (SIM) in the case that the explaning variables are categorical. As a numerical example, we construct a model for the distribution of residential utility by linking SIM to a logit model. The value of this utility can be obtained by using the Wilson's spatial interaction model whose structure is the same as a logit model's. We construct SIM with categorical variables for describing this value.