Students’ academic performance prediction using decision tree

Academic performance is a key indicator of both institutional quality and student development. High dropout rates negatively affect the university’s reputation, making early identification of at-risk students crucial. While many researchers have developed prediction models for academic performance, but this remains a challenge due to the large number of influencing factors. This makes it harder to identify students at-risk and give them effective support. This study develops an accurate predictive model for students' academic performance using a Decision Tree Classification algorithm. This study begins with the preliminary studies and then continues with the data acquisition, data preprocessing, prediction model development and the web-based prototype development. The best model achieved an accuracy of 90.14% using 10-fold cross-validation with a precision of 90.73%, recall of 90.14%, and specificity of 97.74%. These results show the model’s potential to assist educators and administrators in identifying at-risk students and taking timely action to support their academic success.