PREDICTION OF PROSPECTIVE NEW STUDENTS USING DECISION TREE, RANDOM FOREST, AND NAIVE BAYES

Abstract

Higher education positions new student enrollment as a strategic activity for private universities. The effectiveness of selecting prospective students with a high potential to register and be accepted is crucial. Therefore, this study was conducted to find a data classification model that can determine the potential acceptance of new students, allowing private universities to increase the number of students admitted. This research's data originated from the 2020 new student admissions at a prominent private university in Pontianak city. The three chosen classification methods are the decision tree, random forest, and naïve-bayes. Evaluation results indicate the accuracy rate of the decision tree is 59.1%, random forest at 59.2%, and naïve-Bayes at 58.1%. Despite similar accuracy rates, the random forest method slightly outperformed the others, suggesting it may be the most reliable for predicting student enrollment. Based on these models, the estimated potential of prospective students registering at the university ranges from 72% to 78% of the total student candidates. In conclusion, although the three models have almost similar accuracy rates, all show an optimistic estimate regarding the registration potential of prospective students. Thus, universities can use one or a combination of the three models to enhance efficiency in the student admission process.