Modelo de estratificación socioeconómica basado en inteligencia artificial en la localidad de Chapinero, Bogotá D.C.

Abstract

In this work, the accuracy of an artificial intelligence algorithm will be determined and tested through machine learning, such as the Socioeconomic Stratification Model based on Artificial Intelligence for the calculation of the socioeconomic strata of Chapinero locality, through the use of neural networks and/or decision trees. For the work, an alphanumeric database was taken into account along with the facade images of the properties, this for the calculation of the socioeconomic stratification of blocks located in Chapinero locality, in order to minimize the time and money resources of the current methodology and implement the use of new technologies in the territory.

As a first step, images of facades of properties were extracted, which thanks to the extension of the Street View of the company Google offers for free, which together with variables of each of the houses or blocks of the properties under study related to the alphanumeric database that belongs to the Special Administrative Unit of the District Cadastre.

From the WEKA software, the data was processed, where depending on the case of the variable it was converted into nominal or continuous, for the case of the image with .jpg format a different treatment was carried out, subsequently neural networks and decision trees were tested to verify which of these executed algorithms is the best qualifier of socioeconomic stratification for the Chapinero locality.

Among the most optimal classification algorithms is the J48, which will be the Artificial Intelligence Socioeconomic Stratification Model (MESIA) algorithm, since this model manages to define the stratification with more than 90% accuracy, and takes into account very marked limits between strata such as strata 1 and 2 located towards the eastern hills in the Chapinero locality of Bogotá. But what would happen if the Socioeconomic Stratification Model based on Artificial Intelligence is applied in other localities of Bogotá? Or what other variables could be added to the Artificial Intelligence-based Socioeconomic Stratification Model?