Enhancing E-Learning Readiness Assessment Using XGBoost: An Explainable AI Approach

E-Learning readiness assessment in universities has the potential to enhance educational outcomes significantly. This study presents an efficient data-driven approach to enhance the prediction of E-Learning readiness by merging the advantages of the robust XGBoost algorithm and ADKAR factors. To this end, surveys were conducted among faculty members and students at Tlemcen University's Economics faculty in Algeria, focusing on the five critical dimensions of the ADKAR model: Awareness, Desire, Knowledge, Ability, and Reinforcement. Following data validation, XGBoost was trained on a 5-fold cross-validation dataset to prevent overfitting and subsequently used for prediction. Results indicate that the XGBoost model consistently outperforms competing models, including K-Nearest Neighbors, Decision Trees, and Random Forest, achieving an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> of 0.799. A statistical test based on the Bayesian interpretation of the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$t$</tex>-test confirms the consistent superiority of the XGBoost model. Furthermore, feature importance analysis using XGBoost identifies Ability and Knowledge as the primary influencers. Additionally, SHapley Additive exPlanations (SHAP) values underscore Ability as the predominant driver of E-Learning readiness.