MOOC Drop-Out Prediction Based on Learners Sentiment Analysis

Abstract

Despite the progress and appeal of MOOCs as an accessible and inexpensive online learning mode, retention rates remain minimal. There are several reasons for this, including the lack of interaction, the approaches used in course design and presentation, as well as factors related to students' cognitive aspects, their representations, and the obstacles encountered along the way. It is therefore essential to understand students' opinions in the context of MOOCs, using discussion forums as a means of interaction where learners can share their experiences. This paper presents a method for predicting the risk of dropping out of a MOOC by analyzing learners' sentiments in their forum comments. The approach is based on a detailed process of textual data preparation, including cleaning, transliteration, and tokenization. Then, we extract relevant features that feed into a logistic regression model using advanced natural language processing techniques. This predictive model enables us to estimate the risk of dropping out based on student satisfaction, providing a valuable tool for identifying and intervening with at-risk learners.