Predicting Student Dropout in MOOCs Using Genetic Algorithms and XGBoost

Abstract

Massive open online courses (MOOCs) have become a transformative tendency in education due to their various benefits in the last few years. However, despite the various benefits offered to students by MOOCs, such as accessibility, flexibility, and affordability, in addition to the massive number of enrolments, MOOCs suffer from persistent high dropout rates. The emergence of machine learning (ML)and deep learning (DL) techniques, along with educational big data provided by MOOC platforms, allows researchers to address the student dropout problem through Big Data analytics and predictive models to reveal hidden patterns to improve academic outcomes. This study proposes a combination of genetic and XGBoost algorithms for forecasting student dropout from MOOCs as early as possible. It also employed and comprehensively compared several machine learning (ML) and deep learning (DL) predictive models. These models include DT, RF, LR, SVM, MLP, and LSTM. The models suggested in this study to investigate student dropout prediction (SDP) utilize the Open University Learning Analytics Dataset (OULAD). The results showed that the employed models could successfully predict student dropout. However, the proposed model outperforms the other models with 0.36-4.36%,1.34-3.78%, and 1.18-3.20% improvement in terms of accuracy, F1 Score, and AUC, respectively.