Mateo: uma abordagem de descoberta de conhecimento para desvendar as causas da evasão escolar

Abstract

Dropout is one of the most intriguing and crucial problems of education. This problem permeates the various levels and modalities of education and generates social, economic, political, academic and financial losses for all involved in the educational process. Therefore, the development of efficient methods to predict the risk of student dropout is essential, allowing institutions to take proactive actions to minimize the situation. Thus, the objective of this work is to present a knowledge discovery approach, in a database, aiming to identify groups of students at risk of dropping out in face-to-face higher education courses. The approach was validated using data from alumni of the Computer Science college at the Federal Rural University of Semi-Arid, Mossoró/RN. In this work, classification models were used, one of the Artificial Intelligence techniques, which enables continuous learning. In the validation of the approach, five machine learning models were used, and two models obtained better accuracy indexes (SVM and Adaboost). Regarding the results, it was observed that: students who studied in private schools in high school, are more likely to complete the course; as well as the natural students of the state of RN; natural students from other cities who moved to live in Mossoró/RN; the students who lived in the academic village; those with the best monthly student income; students who earn scholarships during the course; students with institutional support; Students who complete the course believe that Information Technology professionals are adequately remunerated and students who have a parent with a college degree.