SCHOLAR AI: INNOVATION IN SCHOLARSHIP SELECTION CLUSTERING BASED ELIGIBILITY CLASSIFICATION BASED ON MACHINE LEARNING

Abstract

Scholarship allocation is a crucial process that ensures financial support for students based on academic performance, potential, and financial need. However, the scholarship selection process at Pondok Pesantren Darunnajah has faced challenges in capturing the holistic characteristics of applicants. This research proposes a machine learning model that integrates clustering and predictive techniques to improve the scholarship selection process. The dataset consists of 300 student samples with attributes such as academic scores, tahfidz (Qur'an memorization), family income, and extracurricular activities. These features help determine if a student qualifies for one of three scholarship schemes: Beasiswa Tahfidz, Beasiswa Prestasi, or Beasiswa Ashabunnajah, or if they are deemed "not eligible." The model follows the Cross-Industry Standard Process for Data Mining (CRISP-DM) framework and utilizes machine learning algorithms for classification. To ensure the model's robustness, its performance is evaluated using K-fold cross-validation, with 5-fold validation employed to validate the model's predictions. The results show a high mean validation accuracy of 90.61% and an F1-score of 0.9311, indicating strong generalization capabilities. These findings highlight the model's potential to improve the scholarship allocation process, ensuring scholarships are awarded to the most deserving students based on academic performance, leadership potential, and financial need. Despite its high performance, the study acknowledges limitations such as potential biases in the dataset and challenges in capturing all relevant factors. These issues may affect the overall effectiveness of the model, suggesting room for improvement in addressing the complexity of the selection process.