Estimating student dropout in distance higher education using semi-supervised techniques

Abstract

Nowadays, distance higher education has rapidly increased due to advance and integration of information and communications' technology. Students who attend online distance courses have often family obligations and job commitments and are usually in 'high risk' of dropout during their attendance. It is of a highly importance to identify such students, through paying extra attention and support to them could possibly minimize the possibility of student failure or even dropout. The present research intends to study whether semi-supervised techniques could be useful in student dropout prediction in distance higher education. Semi-supervised learning aims to generate reliable predictions using few labeled and many unlabeled data. Labeled data are difficult obtainable quite often, as they require many experts, a lot of human effort and time in experiments. As far as, we are aware in several studies propose and compare supervised methods for students' dropout prediction rates in higher education, but none of them investigates the effectiveness of semi-supervised methods. The results of our experiments reveal that a good predictive accuracy can be achieved using few labeled data in comparison to well known supervised learning algorithms. For that purpose we have developed a web-based tool to estimate if an individual student is going to dropout.

Predicting Secondary School Students' Performance Utilizing a Semi-supervised Learning Approach

“…In more recent works, Kostopoulos et al (2015aKostopoulos et al ( , 2015b examined the effectiveness of semisupervised methods for predicting students' performance in distance higher education. Several experiments were conducted using a variety of semisupervised learning algorithms compared with well-known supervised methods which revealed some very promising results, especially the self-training and the tri-training algorithm.…”

“…We refer the reader to the studies by Pise and Kulkarni (2008), Triguero and Garc´ıa (2015), and Zhu (2006) and the references therein for an overview on semisupervised learning methods. During the last decade, many researchers have applied these methods in many real-world applications which have stated that the classification accuracy can be significantly improved if a large number of unlabeled data are used together with a small number of labeled data (Chapelle, Scholkopf, & Zien, 2009;Kostopoulos et al, 2015aKostopoulos et al, , 2015bLevatic et al, 2013;Liu & Yuen, 2011;Sigdel et al, 2014;Triguero, S´aez, Luengo, Garc´ıa, & Herrera, 2014;Wang & Chen, 2013;Zhu, 2006Zhu, , 2011.…”

“…In this study, we investigate the classification accuracy utilizing two wrapperbased semisupervised learning techniques-self-training and YATSI. Selftraining has established as a very popular algorithm due to its simplicity and it has been successfully utilized to tackle difficult real-world problems and is often found to be more efficient and more accurate than other semisupervised algorithms (Kostopoulos et al, 2015a(Kostopoulos et al, , 2015bRoli & Marcialis 2006;Rosenberg et al, 2005;Sigdel et al, 2014). YATSI is a wrapper-supervised algorithm which has been recently proposed in the literature and consists of a two-stage classifier that improves its predictive accuracy by making use of the available unlabeled data, providing some promising results (Catal & Diri, 2009;Levatic et al, 2013;Sigdel et al, 2014).…”

“…Semisupervised learning constitutes the appropriate technique to exploit data originated from educational institutes, since there is often a lack of labeled data, while unlabeled is vast. However, the majority proportion of these studies examines the efficiency of supervised methods, and especially classification (usually pass or fail), while semisupervised learning methodologies have been rarely and only recently applied to the educational field (Kostopoulos et al, 2015a(Kostopoulos et al, , 2015b for predicting students' performance in distance higher education.…”

Implementing AutoML in Educational Data Mining for Prediction Tasks

“…Finally, a recent study explored the usage of semi-supervised techniques for the task of drop out prediction [42]. The dataset consisted of 2 classes of 344 instances characterized by 12 attributes.…”