1-III_pm

[39] New Clustering Method to Estimate Overlapped Exchange Area from Regional Flow Matrix

Characteristics of population migrations and freight flow

○Yudai Honma¹, Atsushi Shirahama (1. Institute of Industrial Science, The University of Tokyo)

Keywords:regional flow matrix, exchenge area, overlapped structure, clustering method, mathematical optimization

In this paper, we proposed a new clustering method to estimate exchange areas from the observed regional flow matrix. In particular, we incorporated not only the average value of intra-flow but also that of inter-flow, and described a overlapped structure of exchange areas. Though our formulation became quadratic and fractional programming, we succeeded in solving the optimal solutions using a solver for constraint programming. We applied the model to population migrations and clarified how the overlapped area increased as time elapsed. We also calculated the results for freight flows and examined how freight types affect the characteristics of exchange areas.