Optimization of Parameter Tuning in the SVM Algorithm Using Metaheuristic Optimization Algorithms for Detecting Stunting Risk in Pregnant Women Based on Urine Test Data

EasyChair Preprint 15846

Abstract

Stunting is a serious health issue affecting children's growth from pregnancy. Early detection of stunting risk in pregnant women is crucial to prevent long-term impacts. This study develops a stunting risk prediction model based on Support Vector Machine (SVM) with parameter optimization using metaheuristic optimization algorithms. The data used is derived from urine test results of pregnant women, encompassing various clinical parameters. The optimization algorithms employed include Grey Wolf Optimizer (GWO), Simulated Annealing (SA), and Firefly Algorithm (FA) to find the optimal C and gamma parameters for SVM. Model evaluation was conducted using accuracy, precision, recall, and F1-score metrics. The results show that optimization with GWO increased the model accuracy to 94.15%, compared to the default model, which only achieved 88.46%. SA optimization also improved accuracy to 94.12%, while FA reached 85.71%. These findings indicate that using metaheuristic optimization in SVM parameter tuning can significantly enhance stunting risk prediction performance.

Keyphrases: Firefly Algorithm (FA), Grey Wolf Optimizer (GWO), Simulated Annealing (SA), Stunting, Support Vector Machine (SVM), parameter tuning

@booklet{EasyChair:15846,
  author    = {Yudha Wibowo and Agung Mulyo Widodo and Gerry Firmansyah and Budi Tjahjono},
  title     = {Optimization of Parameter Tuning in the SVM Algorithm Using Metaheuristic Optimization Algorithms for Detecting Stunting Risk in Pregnant Women Based on Urine Test Data},
  howpublished = {EasyChair Preprint 15846},
  year      = {EasyChair, 2025}}