Article - Main Track

Applying Data Mining for School Dropout Prediction in Higher Education at a Federal Institute for Education

Author: SANTOS, J. C. B., GOULART, A., FEITOSA, S. S. et al.

Abstract: Data Mining is a process that seeks to extract useful knowledge and uncover patterns from data. School dropout is still one of the challenges to be tackled in the Higher Education environment. This paper presents a tool that uses a model that aims to predict a potential dropout of undergraduate students of a higher education institution using Machine Learning algorithms. In order to perform the predictions, we used the Decision Tree and Neural Network techniques, where the former achieved the best performance, with 84% precision and 87% accuracy in detecting dropout, while the second achieved 82% accuracy with 78% precision. Besides, given the data obtained from the institution, the most important features that help prediction school dropout are the average number of classes skipped in previous semester and the student’s age.

Keywords: School Dropout, Data Mining, Classification, Higher Education.

Full paper (in Portuguese)

Full Reference: Santos, J. C. B., Goulart, A., Feitosa, S. S. et al., "Aplicando Mineração de Dados Para Predição da Evasão Escolar no Ensino Superior em uma Instituição Federal de Ensino", Revista de Sistemas de Informação da FSMA n 29(2022) pp. 12-24

Back