Forecasting Student Academic Performance at A Philippine State University Using Supervised Learning Algorithms

Abstract

Strong academic performance is a priority not only for students but also for the institution. Students with good academic standing reflect the academic reputation of universities, serving as an indicator of high-quality education. Consequently, this creates opportunities such as increased funding and improved facilities that benefit students, faculty, and the community. Predicting students' performance in the early stage of the admission process enables institutions to provide tailored support and intervention for students to successfully navigate their academics. This study intends to predict the performance of incoming first-year college students at a state university in Laguna, Philippines, using supervised learning algorithms such as Linear Regression, Random Forest, Artificial Neural Network, and k-nearest Neighbor, employing the Weka software. Furthermore, the study identified the admission criteria with the highest accuracy that can predict student academic achievement. Using a ten-fold cross-validation, results showed that Linear regression is the model that can accurately predict student academic performance in the pre-pandemic dataset with an accuracy rate of 79.63%. The algorithms, however, could not accurately forecast student performance be using the pandemic dataset. Among the criteria used, senior high school GPA (SHS_Grade) is the best predictor for student performance due to its strong correlation with the final grade (Ave_GPA) in both datasets. The findings of this study will help institutions make data-driven decisions on student admission that allow early interventions and enhance educational outcomes.