Optimization of Student Graduation Predictions on Time Using Binning And Synthetic Minority Oversampling Technique (SMOTE)

Abstract

On-time student graduation is a situation where a student graduates from their educational program at the time planned or determined by the relevant educational institution. This research aims to optimize predictions of student graduation on time using the Binning method to group variables into discrete categories and Synthetic Minority Oversampling Technique (SMOTE) to overcome class imbalances in the dataset. Data containing several variables was analyzed using the Naïve Bayes, Decision Tree and Random Forest machine learning algorithms. Model evaluation is carried out using metrics such as precision, Recall, accuracy, and F1-score. The results confirm that the combination of Binning and SMOTE has a significant impact on increasing prediction accuracy. It is hoped that the results of this research can contribute to increasing the accuracy of predicting student graduation on time. By optimizing the use of Binning and SMOTE, it is hoped that the prediction model can overcome the problem of data imbalance and provide more accurate information to higher education institutions to take the necessary preventive actions to increase student graduation rates and become a reference for similar research in the future.