Optimasi Hyperparameter CART Menggunakan Particle Swarm Optimization (PSO) untuk Klasifikasi Penyakit Stroke

Abstract

Stroke is a leading cause of death and disability worldwide, including in Indonesia, making early diagnosis crucial. This study aims to enhance the accuracy of stroke classification using the Classification and Regression Tree (CART) algorithm optimized with Particle Swarm Optimization (PSO). A primary challenge in stroke classification is the prevalence of imbalanced datasets. To address this issue, the hybrid sampling technique SMOTEENN (Synthetic Minority Over-sampling Technique-Edited Nearest Neighbors) was applied to balance the class distribution. The standard CART model (baseline) was first evaluated, achieving an accuracy of 94.41%. Subsequently, PSO was implemented to find the optimal hyperparameter combination for the CART model. The PSO optimization results improved the model's performance; the optimized CART model achieved an accuracy of 94.84%, an increase of 0.43% compared to the baseline model. This improvement demonstrates that the combination of the SMOTEENN method for handling imbalanced data and PSO for hyperparameter optimization is an effective and promising approach to enhance the accuracy of stroke classification models.