Prediction-based techniques on Academic Performance of Graduating Students through Machine Learning

Recent advancements in information technologies have made it a common practice to analyze educational data from various sources. Determining and analyzing these data could improve the education system by identifying the factors influencing students' academic performance and assessing their performance status. Thus, studies on predicting students' academic performance, gaining a high accuracy level, and extracting insights from massive volumes of educational data are still significant factors among researchers. In this study, predicting the academic performance of students using decision tree (DT), logistic regression (LR), random forest (RF), support vector machine (SVM), and naïve Bayes (NB) algorithm utilizing the data from Agusan del Sur State College of Agriculture and Technology (ASSCAT). The result shows that the best model of all five algorithms was the logistic regression, which obtained an accuracy of 0.91, followed by the support vector machine, which yielded an accuracy rate of 0.90. Logistic regression model can assist the university administrators, faculty, and students predict which students may underperform, allowing for timely intervention.