Clasificación de los municipios mediante árboles de decisión en el contexto de la pobreza multidimensional como elemento de optimización de la planificación regional.

Abstract

Poverty in Colombia is a persistent issue that the country has been unable to resolve throughout its history. This problem is exacerbated by deficiencies in planning and territorial management mechanisms, territorial conflicts, difficulty in accessing certain areas, violence, the country's centralist tendencies, and political divisions. These divisions limit the development of territories and the allocation of resources for public investment. The challenges faced by communities are not confined to a single territory but rather function as spatial clusters. Municipal planning, without considering spatial factors and the conditions of neighboring municipalities, restricts the implementation of joint projects to address existing deficiencies. Current planning mechanisms based on indicator analysis do not allow for the creation of regional plans or the identification of these relationships. The advancement of technology, especially in the field of artificial intelligence and its recent applications, offers opportunities for innovation in the areas managed by cadastral and geodesy engineers. Tools like decision trees reduce complexity in decision-making and provide new approaches to solving territorial problems. Along with graphical representations, these tools become a significant aid for regional planning. In this work, we analyze how poverty is spatially related within the country, the centralist tendencies at different scales, the formation of multivariate clusters, and deficiencies in health, housing, education, employment, and childhood. Additionally, we demonstrate the utility of the Midecat methodology for municipal classification and regional planning, where the implementation of a geo-visualizer, along with the generation of graphical outputs at national, regional, and departmental scales, becomes an essential resource for all planning offices across the country.