Predicción de Riesgos Psicosociales en Docentes de Colegios Públicos Colombianos utilizando Técnicas de Inteligencia Artificial

Abstract

This paper presents a new methodology based on machine learning techniques in diagnostics of psychosocial assessments to identify the risk level in teachers of public schools in Colombia. A comparative study of three important models of machine learning for prediction was done: artificial neural networks, decision trees and naive bayes, reducing the dimensionality of the data. This was done by applying genetic algorithms, algorithm of the expected amount of information, the algorithm GainRatioAttributeEval, Pearson's relation coefficient and principal components analysis. A database was used with 5340 epidemiological records, corresponding to psychosocial evaluations of teachers from public schools in the metropolitan area of a Colombian city. The best predictive performance was obtained with the model of artificial neural networks with an accuracy 93%.