Estimativa de desempenho acadêmico de estudantes em um AVA utilizando técnicas de mineração de dados

Abstract

Some educational environments have incorporated software to support or, in some cases, as a basic condition to the availability of courses. In this scenario, stand out Learning Management Systems (LMS) used to support the development of classroom, blended or distance courses. Learning Management System are characterized by storing a large volume of data. However, these environments lack tools to extract useful information for the development of efficient processes for monitoring students’. Thus, this research investigates how data stored in a LMS could be processed to generate information regarding estimates of students’ future academic performance. To obtain this information, first became necessary to select a set of attributes to represent students in an online course using a LMS. This set of attributes was chosen considering three dimensions, selected through the analysis of theoretical bases about online courses: LMS use profile, student-student interaction and bidirectional student-teacher interaction. Applying data mining techniques on the set of selected attributes, it was possible to obtain estimates of students’ future performance. These estimates can support the development of effective processes for monitoring students, activity of fundamental importance in distance learning. In this research, a study with seven experiments were conducted and present different scenarios where estimates of performance can be obtained. The results of these experiments indicate the viability of this proposal, given the promising accuracy rates obtained in the classification of students regarding their final performance in courses.