Proposition and Evaluation of a Territorial Homogeneity Index

  Abstract: The stratification of municipalities in homogeneous areas facilitates the planning and management of space, especially in large areas, since it allows greater coherence in the adoption of public policies and economic incentives. This paper proposes a homogeneity index to evaluate the degree of similarity of territories previously determined by regional public policies. The index works in conjunction with clustering techniques such as TerraSOM and k-means. The proposed index has been applied in the evaluation of 27 Territories of Identity, state of Bahia, created in 2007 by the Superintendence of Economic and Social Studies of Bahia. The data were composed of a set of 45 socioeconomic and land use variables. The analytical evaluation of the proposed index showed that it was simple to implement and interpret, can be applied with any clustering technique, and can be used in different contexts. The results has shown that to the Territories of Identity results have shown similar results for both clustering techniques, but in the presence of outliers, the homogeneity index was less stable when calculated by the k-means technique. There is not statistically significant spatial autocorrelation for the homogeneity index.

  Key words: spatial analysis, spatial dependence, k-means, TerraSOM, regional planning