Evaluating Factors Influencing Learner Satisfaction in Massive Open Online Course Selection: A Data-Driven Approach Using Machine Learning

Massive open online course (MOOC) platforms have seen a significant increase in the number of participants in recent years. MOOCs are at the forefront of an educational environment that has become increasingly computerized, remote, and very competitive, and the competition in the MOOC world is intense. Many factors are influencing the selection of MOOCs by students for online learning. The evaluation of these factors is an important issue which this study aims to assess them through the educational data mining approach. The factors are identified from the learners’ textual reviews in MOOCs. A text mining approach, latent Dirichlet allocation (LDA), has been used to discover the factors impacting learners’ satisfaction in the selection of a MOOC for online courses. Also, the author used filter feature selection and principal component analysis for feature selection and support vector regression (SVR) for predicting learners’ choice preferences in MOOCs. The importance of these criteria was identified by the use of LDA, and the top features were selected by feature selection methods. The SVR technique is to predict learners’ preferences using the full features and selected features of the dataset. The results were presented for different combination of supervised learning techniques and feature selection approaches. The result demonstrated that the combination of support vector regression and filter method (RMSE = 0.3121; R² = 0.8095) has outperformed other method in case of accuracy of predicting learners’ choice preference. The results of this study demonstrated that the proposed methodology is an effective way for MOOCs evaluation for students’ online learning. This study provides a new insight into the evaluation of MOOCs using the proposed method. The research implications are presented for future studies on the development of methods for MOOCs evaluation.