Predicting Timely Students Graduation Using the Decision Tree J48 Method at Universitas Advent Indonesia

Abstract

Every universities wants its students to graduate on time. The reason is that the accreditation of the universities may increase. But in reality, there are students who cannot complete their education on time. Therefore, the author wants to examine how to make a system that can predict whether a student can graduate on time using the J48 decision tree method, so that actions can be taken to help the student finishes their studies on time if they tend not to graduate on time. Through the interviews and literature studies, 19 attributes were found that could affect the timeliness of the students graduation at Universitas Advent Indonesia. To process the data, the author uses the WEKA application by implementing the J48 algorithm.

     Results of the study has obtained an accuracy of 90,24%, it can be concluded that this algorithm has a fairly good level of accuracy to make predictions of students graduating on time. The results of the study have concluded that academic leave is main reason for the students not to graduate on time.  While passing the classes, environmental influences, financial difficulties, discipline problems, perseverance in doing assignments, working as student labor in the university, the linearity of studies in accordance with high school specialization, numbers of students’ organizations they followed, and the friendship influences are supporting attributes that can be used to predict.

Keywords: Data Mining, Decision Tree, Timely Prediction