Análise de dados acadêmicos e construção de plataforma para apoio a redução da retenção no ensino superior

Abstract

Retention in higher education is a complex phenomenon that causes major impacts on society, due to all the factors that are intrinsically related to its occurrence, such as financial, educational and social. This monograph has retention as the object of study and presents an analysis of academic data using data mining techniques and the steps of building a platform to make the analysis more accessible. Moreover, automatic mechanisms for communication with students and devices are implemented so that teachers can extract insights and take actions in order to contribute to the reduction of academic retention. With this work it was possible to check the partial influence of some variables on the students' performance, such as the frequency and the performance of the activities suggested by the teachers. In addition, machine learning models were trained considering different scenarios, and among the algorithms used, for the available data set the Random Forest algorithm presented the best results. For the most common scenarios, accuracy between 64.5% and 95.4% was obtained, depending on the time of the school year in which the predictions were made.