<![CDATA[Students are a major part of the life cycle of a university. The number of students graduating from a university often has a small ratio when compared to the number of students obtained in the same academic year. This small student graduation rate can be caused by several aspects, such as the many student activities accompanied by economic aspects, as well as other aspects. This makes it mandatory for a university to have a model that can take into account whether the student can graduate on time or not. One of the main factors that determine the reputation of a university is student graduation on time. The higher the level of new students at a university, with the same ratio, there must also be students who graduate on time. An increase in the number of student data and academic data occurs if many students do not graduate on time from all registered students. So that it will affect the image and reputation of the university which can later threaten the accreditation value of the university. To overcome this, we need a model that can predict student graduation so that it can be used as policy making later. The purpose of this study is to propose the best classification model by comparing the highest level of accuracy of several classification algorithms including Naïve Bayes, Random Forest, Decision Tree, K-Nearest Neighbor (K-NN) and Support Vector Machine (SVM) to predict student graduation. In addition, the feature selection process is also used before the classification process to optimize the model. The use of feature selection in this model with the best features using 12 regular attribute features and 1 attribute as a label. It was found that the classification model using the Random Forest algorithm was chosen, with the highest accuracy value reaching 77.35% better than other algorithms.]]>
