Estudo sobre docentes do ensino básico através de indicadores educacionais e modelos de regressão

Abstract

This work uses educational databases provided by INEP and applies regression techniques with the purpose of estimating the indicator of teachers with higher education in the scenario of basic education. This application of Educational Data Mining is based on the phases of the CRISP-DM methodology. After correlational analyzes of the variables, linear and robust regression models were applied in order to compare the performance and provide a model that minimizes the prediction error. The models were evaluated by absolute mean error in addition to graphs and statistical tests. The results indicate that the robust regression obtained better results in the estimation of the variable listed in this study.