Determining Prospective Recipients of The Indonesia Smart Scholarship Program Using The Support Vector Machine Algorithm and Simple Additive Weighting

Abstract

Between the period from 2018 to the first semester of 2020, a total of 2.4 million students who owned the Kartu Indonesia Pintar (Indonesia Smart Card) were at risk of not being able to access the Program Indonesia Pintar due to delays in entering recipient data, misdirected targeting, and fund disbursement obstacles. During its implementation, there were issues with inaccurate data of Kartu Indonesia Pintar recipients, resulting in the beneficiaries not being entirely targeted correctly. According to Article 12 of Minister of Education and Culture Regulation No. 10 of 2020, schools, as educational units, are assigned the responsibility of managing the Smart Indonesia Program at the school level. This includes tasks such as proposing eligible student candidates for the Program based on criteria, monitoring and assisting the smooth process of receiving the benefits, and accommodating students who hold the Kartu Indonesia Pintar (KIP). However, schools faced difficulties in obtaining the eligibility data of potential Indonesia Pintar Program recipients through the school proposal format. The objective of this research is to determine the eligibility of potential Indonesia Pintar Program recipients using a combination of Support Vector Machine (SVM) and Simple Additive Weighting (SAW) techniques. The SVM model was trained using kernel tricks and parameter selection (C and gamma) based on data sourced from the DAPODIK application of students in SMK Negeri 1 Kota Bekasi for the academic years 2020/2021 to 2022/2023, with a dataset containing 2499 rows . This dataset was used to determine the eligibility of potential Indonesia Pintar Program recipients. The SAW method was then employed to calculate scores based on the weighting of attributes possessed by the students, which were subsequently ranked from the highest score. The Linear Kernel with parameter values C=0.1 and gamma=0.1 achieved an accuracy of 88.8% during the classification of test data, using 10% of the total dataset, which amounted to 250 rows of data. The combination of SVM and SAW methods resulted in the classification of eligibility for potential Indonesia Pintar Program recipients, ranked based on their scores using the weighted assessment with the SAW method.