Volume 5, Issue 2 (2025) – 4 articles
Cover Picture: This study presents an innovative multimodal machine learning and explainable AI (XAI) framework to predict and interpret the impact of ChatGPT usage on students’ academic satisfaction and performance. By integrating XGBoost, Random Forest, and Support Vector Machine models with SHAP and LIME explanation techniques, the approach ensures both high prediction accuracy and transparency in model decision-making. Leveraging a custom-built ChatGPT Survey dataset, the results show that Support Vector Machine achieves 89% accuracy (AUC-ROC: 92%), while XGBoost reaches the highest accuracy (92%) in regression analysis. XAI analysis highlights ChatGPT usage and satisfaction as the most influential factors, offering new insights into responsible AI application in educational settings.
view this paper 
