Indonesia Conference Directory

The purpose of this study is to predict students who have the potential to drop out in a college so that the prospective student selection process is more effective. Based on the problems that have been raised, the forecasting method is proposed to predict prospective students to drop out before entering the lecture using the Decision Tree C4.5 and Forward Selection methods. The tool used in this study uses rapidminer 9.2, the results obtained using 90% training data and 10% testing data resulted in an accuracy of 82.52% and obtained attribute models that influence student graduation classifications, namely the attributes of the Study Programs and Age.