A Comparative Study of Machine Learning Algorithms for Regression in Predicting the Academic Performance of Students in General Mathematics

Abstract

This study explores the application of predictive modeling techniques in assessing the academic performance of Senior High School students at Notre Dame of Midsayap College, focusing on General Mathematics. Employing three distinct machine learning algorithms — multiple linear regression (MLR), random forest regression (RFR), and support vector regression (SVR) — the study aims to predict students’ General Mathematics grades. Evaluation of these algorithms’ predictive capabilities is conducted utilizing metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and adjusted R-squared. Results indicate that the multiple linear regression model exhibits superior predictive performance, yielding lower RMSE and MAE values compared to RFR and SVR models, achieving an accuracy prediction of 97.29%.