Abstract
Academic performance depends on a multivariable nexus of socio-academic andfinancial factors. This study investigates these influences to developeffective strategies for optimizing students' CGPA. To achieve this, wereviewed various literature to identify key influencing factors and constructedan initial hypothetical causal graph based on the findings. Additionally, anonline survey was conducted, where 1,050 students participated, providingcomprehensive data for analysis. Rigorous data preprocessing techniques,including cleaning and visualization, ensured data quality before analysis.Causal analysis validated the relationships among variables, offering deeperinsights into their direct and indirect effects on CGPA. Regression models wereimplemented for CGPA prediction, while classification models categorizedstudents based on performance levels. Ridge Regression demonstrated strongpredictive accuracy, achieving a Mean Absolute Error of 0.12 and a Mean SquaredError of 0.023. Random Forest outperformed in classification, attaining anF1-score near perfection and an accuracy of 98.68%. Explainable AI techniquessuch as SHAP, LIME, and Interpret enhanced model interpretability, highlightingcritical factors such as study hours, scholarships, parental education, andprior academic performance. The study culminated in the development of aweb-based application that provides students with personalized insights,allowing them to predict academic performance, identify areas for improvement,and make informed decisions to enhance their outcomes.