Robust Principal Component Regression with Wild Bootstrap for Handling Outliers, Multicollinearity, and Heteroskedasticity in Chronic Hepatitis B Data

This study aims to analyze the performance of the Robust Principal Component (RPC) method combined with Wild Bootstrap in handling outliers, multicollinearity, and heteroskedasticity in chronic hepatitis B data. The data were obtained from the World Health Organization (WHO) and consist of several epidemiological indicators. The analysis methods include Principal Component Analysis (PCA), Least Trimmed Squares (LTS), and Wild Bootstrap using Wu and Liu multipliers. The results show that the dataset contains outliers, strong multicollinearity, and heteroskedasticity. The RPC-Wild Bootstrap method produces more stable parameter estimates, with RPC Boot Wu showing lower standard error and RMSE compared to RPC Boot Liu. Therefore, the RPC-Wild Bootstrap method is effective in producing more stable and reliable parameter estimates for complex real-world data.