Talent Identification System Using Support Vector Machine (SVM) for Perlis Sports School

The Talent Identification System is a machine-learning-based system that makes suitable sports recommendations to students. This study develops a Sport Talent Identification System using a Support Vector Machine (SVM) trained on physical and physiological performance data collected from the Sports Council (MSN) of Perlis athletes. To improve prediction reliability and reduce class imbalance, the dataset was grouped into five major categories: ‘Sukan Ke- cepatan/Kekuatan’ (Power), ‘Sukan Combat’, ‘Sukan Berkumpulan’, ‘Sukan Raket’, and ‘Sukan Skill’. The results show that the SVM model achieved 63.03% test accuracy, outperforming KNN. This approach aims to improve the talent identification process in sports schools, provide accurate and reliable sports recommendations, and enable coaches to make informed decisions about the student-athlete.