SYSDAE: usando Machine Learning numa análise do perfil comportamental dos alunos do ensino médio do IFRS - Campus Sertão

Abstract

This work aimed to predict the risk of losing a student's residence high school students, employing artificial intelligence, which was used in the area of Machine Learning, machine learning algorithms. The analysis is based in various data related to student residence, including information about the student, course, number of hours taught, measure, sector, number of days suspended, gender, grade and enrollment to generate a predictive system. The risk of losing student residence consists of the fact that the student can no longer remain as residing close to the campus, causing several difficulties, such as living outside the limits of the institution, having to pay rent, a reality not considered for many students from our institution. The act of remaining resident is a reflection of the The student's pattern of behavior is seen as the result of the model. Three Different machine learning algorithms were considered for identify and classify the parameters that affect permanence in the residence student: Naive-Bayes, KNN, Decision Tree. To evaluate the performance of machine learning algorithms, three metrics were used: accuracy, recall and F1-score. The results indicate that KNN outperformed other techniques, generating superior results, followed by Naive-Bayes. It is concluded that there is a way to predict the risk of loss of student residence based on a standard of student behavior. A web application was developed for the presentation of results. According to the experiment carried out, the approach proved to be adequate and can serve as decision-making in actions aimed at better relationship between Student Residence and students, in addition to improving students' social skills, necessary for socializing among peers.