Identificación de clusters espaciales de empresas y la influencia de factores externos en su constitución utilizando técnicas de minería de datos.

Abstract

This work focuses on the use of different data mining techniques, based on geospatial statistical methods for the identification of patterns with respect to the economic activities of companies, registered in the Superintendencia de Compañías in the Metropolitan District of Quito, at as well as defining external factors that influence the constitution of these. First, to create the clusters, the Local Indicators of Spatial Association (LISA) are used, which will define the neighborhoods that have high density and are characterized with auxiliary variables that describe the location. This also helps to identify potential neighborhoods with similar characteristics, but without a high density of businesses. Then, spatial regression models are used to identify the relationship that exists between the companies and the auxiliary variables found, analyzing the coefficient associated with each one. Finally, by complementing both results, the list of high-density neighborhoods in which work should be done is obtained, with the factor of growth or decrease of companies for each of the auxiliary variables identified.