Optimasi Hyperparameter CatBoost dengan Particle Swarm Optimization untuk Klasifikasi Hipertensi

Muhammad Iqbal Al Afgany (Universitas Pembangunan Nasional “Veteran” Jawa Timur, Jl. Rungkut Madya No. 1, Gn. Anyar, Surabaya, Indonesia)
Ani Dijah Rahajoe (Universitas Pembangunan Nasional “Veteran” Jawa Timur, Jl. Rungkut Madya No. 1, Gn. Anyar, Surabaya, Indonesia)
Henni Endah Wahanani (Universitas Pembangunan Nasional “Veteran” Jawa Timur, Jl. Rungkut Madya No. 1, Gn. Anyar, Surabaya, Indonesia)

JASIEK (Jurnal Aplikasi Sains, Informasi, Elektronika dan Komputer)
Vol. 7 No. 2 (2025): Desember 2025

Hypertension is a cardiovascular disease affecting 11,952,694 residents aged ≥15 years in East Java in 2019, yet only 40.1% received healthcare services. This study aims to analyze the effect of Particle Swarm Optimization (PSO) on CatBoost algorithm performance in hypertension level classification. The research dataset combined data from Puskesmas Kepatihan Gresik (191 data) and Kaggle (12,500 data) divided with an 80:10:10 ratio. PSO was used for CatBoost hyperparameter optimization including iterations, depth, learning_rate, and l2_leaf_reg. Model evaluation utilized accuracy, precision, recall, and F1-score metrics. Results show that CatBoost with PSO optimization achieved 96% accuracy with optimal configuration of iterations=100, depth=3, learning_rate=0.055, and l2_leaf_reg=3, 2% higher than without optimization (94%). This study proves the effectiveness of PSO in optimizing CatBoost hyperparameters for more accurate early hypertension detection