Prediksi Kelulusan Tepat Waktu Menggunakan Metode C4.5 DAN K-NN (Studi Kasus : Mahasiswa Program Studi S1 Ilmu Farmasi, Fakultas Farmasi, Universitas Muhammadiyah Purwokerto)

Abstract

The graduation profile is one of the key elements for the accreditation standard of higher education. It mirrors the performance of the applied educational system within a period of time. The better it is, the better the accreditation will be. In support of this, a graduation prediction may be conducted to the academic database of the students. It is of pivotal to trace and classify the historical data into the data training and data testing, thus, to predict the on time-graduation. The step is importantly done to help decide the better management of learning processes. This study was therefore done to analyse certain variables applied to predict the on time-graduation using the algorythms of C.45 and K-Nearest Neighbour (K-NN). The data mining was done to the academic database of the students of the Pharmacy study programme, Pharmacy Faculty, Muhammadiyah University of Purwokerto by adding certain variables into the process. The data was then classified into the data training and data testing. Backward selection was done to select the best and most influential variables for the dataset. The study further resulted that by using the algorhythm of C.45 and backward selection, the accuracy of the graduation reached 92.75%. It is different from the acurracy the K-NN and backward selection showed that reached 96.14%. The result confirmed that the KNN showed the better accuracy than the C.45. It considerably benefitted the study programme to make better decisions on increasing the quality of services, in particular that of leraning processes.