Machine Learning-Based Classification For Selection Of Smart Scholarship Fee Amounts In The Capital City

In the management of scholarship funds such as the Jakarta Smart Card (KJP), challenges related to accuracy and efficiency are often faced, so the application of machine learning methods is expected to improve the results of data processing of scholarship recipients. This research aims to improve the accuracy of the KJP scholarship fund receipt process by applying a machine learning-based classification method. The KJP Scholarship has an important role in supporting education in Jakarta, but there is a need to improve the distribution process and the accuracy of fund distribution. By utilizing machine learning techniques, this research focuses on processing scholarship recipient data to produce better decisions. This study uses a quantitative approach with a data analysis method, with the population of all KJP scholarship recipients in 2023 and a random sample taken from the available data. Data was collected through documentation and processing of historical data of scholarship recipients, with analysis using the SEMMA (Sample, Explore, Modify, Model, Assess) approach as well as Decision Tree, Naïve Bayes, and Random Forest algorithms. The results of the analysis show that the Decision Tree model provides the best performance with an accuracy of 88.31%, compared to Naïve Bayes and Random Forest. These findings provide new insights for better decision-making in the distribution of KJP scholarships, as well as demonstrate the potential for the integration of machine learning in scholarship management in the education sector, which can improve the efficiency and accuracy of fund management.