An efficient use of ensemble methods to predict students academic performance

Authors:
Publisher website
Google Scholar
Abstract
Application of data mining techniques in an educational background can discover hidden knowledge and patterns that will support in decision-making processes for improving the educational system. In e-learning system or web-based education, student's behavioral(SB) features play an important role that will show the student's interactivity towards the e-learning system. The aim of this paper is to show the importance of SB features and for this task we have collected the educational dataset from learning management system (LMS). On the included dataset, feature analysis has been done and after that, we have used data preprocessing phase that is an important step in knowledge discovery process. On the preprocessed dataset, classification is performed on it by using classifiers namely; Decision Tree (ID3), Nave Bayes, K-Nearest Neighbor, Support vector machines to predict student's academic performance. The accuracy of the proposed model is achieved by using Ensemble Methods. We have used Bagging, Boosting, and Voting Algorithm that are the common ensemble methods. On using ensemble methods, we have got the better result that proves the reliability of the proposed model.

This publication has 7 references indexed in Scilit: