Predicting Student's Success in Programming Courses: A Decision Support System for Admission in Computer Science and Information Technology Programs

Predicting students' success prior to their admission to computer-related degree programs is challenging due to diverse educational backgrounds and varying skill requirements. Educational Data Mining (EDM) can be used to optimize the college admission process by selecting only well-equipped students for the program. The attrition rate of the College of Computing Studies at Western Mindanao State University remains high despite the rigorous admission process for the Computer Science and Information Technology programs. Many students fail programming courses, causing them to leave the program. This study developed a decision support system and ensemble models by combining Support Vector Machine, Decision Tree, and Neural Network models using learning techniques such as voting, bagging, and stacking. The goal is to automate the admission of pre-qualified students based on their likelihood of success in programming courses. Factors such as SHSGPA, CET score, class rank, and specific personality traits significantly influence academic success. The combined models, particularly using stacking techniques, emerged as the most reliable and effective model for predicting unseen data compared to standalone classification models. Students and faculty members evaluated the functionality, reliability, and usability of the decision-support system. The feedback indicated a positive reception, affirming its potential as a valuable tool in the college's admission process. Future enhancements may include incorporating multiple intelligence and grit tests and exploring additional ensemble methods to improve predictive capabilities.