Análisis de efectividad al implementar la técnica de árboles de decisión del enfoque de aprendizaje de máquina para la determinación de avalúos masivos para las UPZ 79 Calandaima, 65 Arborizadora y 73 Garcés Navas

Abstract

The present project aims to use the decision tree method, the machine learning approach, within the process that constitute the mass valuations and analyze its effectiveness with respect to the traditional method of linear regression, to make such a comparison, a classification Of datasets of zones of the city of Bogotá D.C., corresponding to the ZPU (Zonal Planning Units) 79 Calandaima, 65 Arborizadora and 73 Garces Navas, making use of the classification methods ID3, J48 and M5P, after which Are evaluated by means of the Cross Validation and Percentage Split tests, obtaining as a result that the decision tree tool is useful and effective in the process of performing mass valuations, presents results closer to the observed values and allows a clear understanding Each rule generated.