Classification Over Geographical Zones: A Combinatorial Optimization Approach to the Regional Partitioning Problem

The regional partitioning problem consists in determining an aggregation of geographical areas into a predefined number of homogeneous and spatially continuous regions. A common difficulty that stems from this problem is the multiple comparisons required by the chosen classification process, which makes it a combinatorial problem. In this sense, the application of heuristic methods becomes necessary to solve it. This work presents the implementation of a method of simulated annealing that minimizes the distances from the objects to the nearest centroid, within the PAM classification method (Kaufman and Rousseeuw, 1987) as applied to the geographical units of the Metropolitan Zone of Mexico's Valley.