Penerapan Teknik Bagging Berbasis Naïve Bayes untuk Seleksi Penerimaan Mahasiswa

Students who graduate not on time create an imbalanced ratio between lecturers and students. The current selection system is ineffective because it has not been able to detect prospective students who have the possibility of not being able to complete their education on time so that many students who are accepted do not graduate on time and leave without completing their education. This condition causes a decrease in performance of study programs and institutions. The classification algorithm can use for classifying new students as graduate timely or not. Naive Bayes classification algorithm can use to classify data in certain classes, using the history of alumni of informatics engineering at Pamulang university as training data and prospective student data as test data. Some attributes used to determine which label class to graduate on time and not on time are gender, school majors, year difference, math grades, English, Indonesian. To improve the results of the classification of Naive Bayes, Bagging (Bootstrap Aggregating) technique is used. From the test results of the alumni dataset, the informatics study program using bagging techniques as an optimization of the Naive Bayes classification algorithm has a lower failure rate than without using bagging techniques. The results of the calculation of performance data using bagging techniques can increase accuracy by 2.381% and AUC by 1.470% on the student graduation prediction model for new student selection using the Naive Bayes classification.