Improving Decision Support Systems in Education Systems Using Data Mining and Machine Learning Techniques

Authors

  • Kinda Abu Qasim Tishreen University
  • Lama Basha Tishreen University

Keywords:

: Ensemble, Voting Method, students' levels, WEKA, Clustering Algorithm

Abstract

Educational data mining aims to study the available data in the educational field and extract the hidden knowledge from it in order to benefit from this knowledge in enhancing the education process and making successful decisions that will improve the student’s academic performance.

This study proposes the use of data mining techniques to improve student performance prediction. Three classification algorithms (Naïve Bayes,J48, Support Vector Machine) were applied to the student performance database, and then a new classifier was  designed to combine the results of those individual classifiers using Voting Method.

The WEKA tool was used, which supports a lot of data mining algorithms and methods.

The results show that the ensemble classifier has the highest accuracy for predicting students' levels compared to other classifiers, as it has achieved a recognition accuracy of 74.8084%.

The simple k-means clustering algorithm was useful in grouping similar students into separate groups, thus understanding the characteristics of each group, which helps to lead and direct each group separately.

Author Biographies

Kinda Abu Qasim, Tishreen University

Associate Professor, computer and automatic control engineering, faculty of Mechanical and electrical engineering

Lama Basha, Tishreen University

Postgraduate Student (Master degree) , Department of computer and automatic control Engineering, Faculty of Mechanical and Electrical Engineering,

Published

2021-11-07

How to Cite

1.
أبو قاسم ك, باشا ل. Improving Decision Support Systems in Education Systems Using Data Mining and Machine Learning Techniques. Tuj-eng [Internet]. 2021Nov.7 [cited 2024Nov.23];43(5). Available from: https://journal.tishreen.edu.sy/index.php/engscnc/article/view/11042

Most read articles by the same author(s)