Machine Learning in Academic Performance Prediction: Analyzing Attendance and Marks to Forecast Future Results

In the realm of higher education, predicting student performance is crucial for enhancing educational outcomes and institutional efficiency. This project aims to develop a predictive model to forecast students upcoming semester performance based on their past academic records and attendance. This paper makes use of machine learning regression techniques, including Random Forest, Gradient Boosting, and Voting Regressor, to analyze students' scores and attendance data. The dataset comprises marks and attendance records for analytical, theoretical, and practical subjects, with target grades categorized into seven classes: A+, A, B+, B, C, P and F. To evaluate model performance, continuous predictions were discretized into these grade categories, allowing for the calculation of Precision, Recall, F1-Score, and the generation of confusion matrices. The results indicate the efficacy of different models in accurately predicting student grades. Visualizations such as bar charts and heatmaps were used to present and compare the performance metrics across various models and subject types. The insights from this project are expected to assist educators in identifying at-risk students and improving educational strategies, ultimately contributing to enhanced student success and institutional reputation.