Educational Data Mining Applied to a Massive Course

Abstract

In the current context of distance learning, learning management systems (LMSs) make it possible to store large volumes of data on web browsing and completed assignments. To understand student behavior patterns in this type of environment, educators and managers must rethink conventional approaches to the analysis of these data and use appropriate computational solutions, such as educational data mining (EDM). Previous studies have tested the application of EDM on small datasets. The main contribution of the present study is the application of EDM algorithms and the analysis of the results in a massive course delivered by a Brazilian University to 181,677 undergraduate students enrolled in different fields. The use of key algorithms in educational contexts, such as decision trees and clustering, can reveal relevant knowledge, including the attribute type that most significantly contributes to passing a course and the behavior patterns of groups of students who fail.