Comparación de la validez y efectividad de gradient boosting y random forest en la estimación de zonas de valor de terrenos rurales en los corregimientos de san sebastián de palmitas y San Cristóbal, Medellín, Antioquia

Abstract

This study aim to compare the validity and effectiveness of the artificial intelligence algorithms Gradient Boosting and Random Forest for estimating the value zones of purely rural lands in the districts of San Sebastián de Palmitas, commune 50 and San Cristóbal, commune 60 located in the rural area of ​​the city of Medellín, Antioquia. The research carried out is of an applied type, with an experimental design, carried out with a sample of 570 purely rural properties of the selected districts in the city of Medellín. The main data collection technique was the consultation of secondary sources available for free use by the Mayor's Office of the city of Medellín. It is concluded that cross-validation and graphical tests confirm the effectiveness of artificial intelligence algorithms as a tool for the constitution of value zones. In the same way, it is pertinent to say that in the case of searching for more precise but less exact values ​​in traditional rural areas, the Random Forest algorithm seems to have a better performance, while in the case of wanting to increase accuracy in technologically advanced or unconventional rural areas, it is the Gradient Boosting algorithm that represents a much more suitable tool.