Comparison of Different Classification Techniques to Predict Student Graduation

Abstract

Every year, the number of students accepted at the Maulana Malik Ibrahim State Islamic University of Malang continues to increase. Still, not all students can graduate on time according to the specified study period, resulting in a buildup of students who have not graduated according to their graduation period. One of the aspects evaluated in the Study Program accreditation process is the student graduation rate. Apart from that, for each semester, Study Programs are also required to report educational data to DIKTI, and student graduation is one of the factors considered in the report. There is an imbalance between the number of students graduating each year and the number of new students accepted. To overcome this problem, it is necessary to predict student graduation to determine whether they will graduate on time. In science and data analysis, predictions are often used to make predictions based on existing data and information. Classification models in predicting student graduation include the Nave Bayes method, Support Vector Machine SVM, and Random Forest, as well as the level of accuracy of these three methods. From the results of experiments and model evaluations carried out, with data from 458 Informatics Engineering Study Program students with details of 366 training data and 92 testing data, it was obtained that the SVM model had the highest accuracy, reaching around 87% and Random Forest also had good accuracy, around 82%. At the same time, the Naïve Bayes model has lower accuracy, around 76%.