Classification Of KIP-K Scholarship Using Logistic Regression, Classification Trees, and Boosting Based On Decision Support System

Abstract

This study addresses the challenge of accurately identifying eligible awardees of the KIP-K scholarship at UIN Sjech M. Djamil Bukittinggi, where scholarship aid requests exceed the allocated funds. The research aims to develop an integrated classification and decision-making model to optimize the selection process. From the 2022 and 2023 scholarship applicant data obtained through AKAMA, preprocessing was conducted, resulting in a final dataset comprising 2,144 records. The dataset includes 14 explanatory variables influencing scholarship eligibility. The study compares three classification methods—logistic regression, classification tree, and boosting—using the 2022 data for training and testing. The SMOTE resampling technique was applied to address class imbalance. The novelty of this research lies in integrating classification analysis with a decision-making system based on the Simple Additive Weighting (SAW) method, enhancing the ranking of applicants based on criteria. The results indicate that logistic regression delivered the best performance in terms of accuracy, sensitivity, and AUC-ROC scores during testing, despite a slight decline in performance when applied to the 2023 dataset. Moreover, integrating logistic regression with SAW significantly improved decision-making precision. The application of logistic regression combined with SAW on the 2023 dataset resulted in a final accuracy of 0.5734 and a balanced accuracy of 0.5820. This integrated framework provides a data-driven, fair, and efficient approach to scholarship allocation. The study highlights the importance of combining predictive models with decision-making systems to ensure equitable and targeted distribution of financial aid to deserving students.