Predicting At-Risk Students’ Performance Based on LMS Activity using Deep Learning

It is of great importance for Higher Education (HE) institutions to continuously work on detecting at-risk students based their performance during academic journey with the purpose supporting success and advancement. This where Learning Analytics (LA) representing learners’ behaviour inside Management Systems (LMS), Educational Data Mining (EDM), Deep (DL) techniques come into play as an sustainable pipeline, which can be used extract meaningful predictions future online activity. Thus, aim this study implement a supervised learning approach utilizes three artifcial neural networks (vRNN, LSTM, GRU) develop models that classify students’ final grade Pass or Fail number LMS activity indicators; more precisely, detect failed who are actually ones susceptible risk. The alongside baseline MLP classifier have been trained two datasets (ELIA 101- 1, ELIA 101-2) illustrating assessment 3529 enrolled in English Foundation-Year course 101) taught at King Abdulaziz University (KAU) first second semesters 2021. Results indicate though all DL performed better than baseline, GRU model achieved highest classification accuracy both datasets: 93.65% 98.90%, respectively. As regards predicting students, = 81% Recall values, notable variation depending dataset, being 101-2.