Uso da mineração de dados na identificação de alunos com perfil de evasão do ensino superior.

Abstract

Reduce students dropout rates in graduate courses is a challenge faced by universities. This problem occurs in public and private institutions, and its effects are related to financial issues and the declining number of graduates in higher education. This paper presents a study on dropout in higher education and the application of data mining techniques. The aim is to find students with evasive trends and the patterns associated with these trends. As part of the theoretical background, were found the main concepts related to the topic, survey related work, study of techniques and algorithms for data mining, and as a methodology for the study, we chose the data mining tool WEKA. In addition to the known mining techniques such as classification and association, to conduct experiments according to the field of data provided by the University of Santa Cruz do Sul - UNISC. These data were given by the university's IT department in text file format and was imported into a supporting database, then transformed into .arff files required by WEKA software and then subjected to data mining algorithms. The results show that the final status of disciplines in the first semester and the number of disciplines taken semester by semester are the factors that collaborate to students dropout.