Preditor de desempenho de estudantes do curso de Matemática Industrial/UFG via redes neurais MLP e SOM

Abstract

In this work, Educational Data Mining techniques are used based on the combination of supervised and unsupervised machine learning methods by Multilayer Perceptron-type Artificial Neural Networks. (MLP) and Auto-Maps organizers of Kohonen (SOM), respectively, for the development of a tool capable of classifying the academic performance of students based on grades in subjects in the on-site course of Bachelor of Industrial Mathematics. The tool developed for classifying students, with the data used in this work, accurately obtained the prediction of academic performance, enabling course managers to use it as an aiding method for decision-making.