Mining Educational Data to Predict Students Dropout in a Professional Public School Institution

Abstract

School data management received technological support and is an important management tool, both from an educational and administrative perspective. Educationally, these data can help to identify students' dropout problems and contribute to the identification of possible profiles for it. Thus, this work aims to collect data and mine for the application of machine learning at IFSP, with special attention to the Itapetininga campus, to verify whether it is possible to predict profiles of students with the possibility of school dropout. Therefore, this work is justified in terms of the attempt to maximize the use of public resources with a reduction in school dropouts and to enhance the success of students on campus. The results of the work carried out showed, with a certain margin of safety, the prediction of school dropout with the data considered from the pre-processing SUAP database, followed by machine learning techniques application and converging to the data visualization results herein presented. Henceforth, it provides a computational tool to support school management and, mainly, encourage practices that seek to reduce school dropout on the Itapetininga campus.