Robust Weighted Least Squares Estimations Against Outliers and Heteroscedasticity

This project evaluates the performance of robust regression methods—WLS, Huber, Tukey, and LTS—against OLS under simulated heteroscedasticity and outlier conditions. While OLS and WLS break down in the presence of outliers, Huber, LTS, and especially Tukey Bisquare estimator shows much stronger robustness and resilient to data contamination.