Title: Forecasting students' success in an open university

Authors: George Kostopoulos; Sotiris Kotsiantis; Christos Pierrakeas; Giannis Koutsonikos; George A. Gravvanis

Addresses: Educational Software Development Laboratory (ESDLab), Department of Mathematics, University of Patras, Greece ' Educational Software Development Laboratory (ESDLab), Department of Mathematics, University of Patras, Greece; Hellenic Open University, Parodos Aristotelous 18, 26335 Patras, Greece ' Technological Educational Institute of Western Greece, Meg. Alexandrou 1, 26334 Patras, Greece; Hellenic Open University, Parodos Aristotelous 18, 26335 Patras, Greece ' Technological Educational Institute of Western Greece, Meg. Alexandrou 1, 26334 Patras, Greece ' Department of Electrical and Computer Engineering, School of Engineering, Democritus University of Thrace, University Campus, Greece; Hellenic Open University, Parodos Aristotelous 18, 26335 Patras, Greece

Abstract: In recent years, students' performance prediction has been identified as one of the most essential and challenging research topics for educational institutions. The necessity for exploitation and analysis of data originating from several educational contexts has led to a widespread implementation of familiar machine learning methods trying to effectively analyse students' academic behaviour and predict their performance. The early detection of low performers is of major importance for open universities seeking to decrease dropout ratios, improve educational outcomes and provide high quality education. This paper introduces an ensemble of classification and regression algorithms for predicting students' performance in a distance web-based course. Several state-of-the-art machine learning methods have also been applied to compare the efficiency of our method. A plethora of experiments have been conducted for this purpose, using data provided by the Hellenic Open University. The proposed ensemble combines classification and regression rules and is as accurate as the powerful ensembles, while the produced model remains comprehensive. In addition, a prototype software support tool has been designed and it simulates the presented ensemble.

Keywords: educational data mining; EDM; performance prediction; classification; regression; ensemble; distance learning; learning technology; open university.

DOI: 10.1504/IJLT.2018.091630

International Journal of Learning Technology, 2018 Vol.13 No.1, pp.26 - 43

Received: 08 May 2021
Accepted: 12 May 2021

Published online: 27 Apr 2018 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article