Best Robust to Estimate The Parameters Using Hampel Weighted Ridge Least Trimmed Squares in Presence of Multicollinearity And Outliers

Abstract

The production of the undesirable impacts on the least squares estimators is by multicollinearity and outliers which are considered as problems in multi regression models. In the current study, an experimental comparative investigation is made for diverse estimation methods. which namely the Ordinary Least Squares (LS), Ridge Regression (RID), Hampel Weighted Ridge Least Absolute Value (HRLAV) and Hampel Weighted Ridge Least Trimmed Squares (HRLTS). From a numerical example and a simulation study, the resulting Hampel Weighted Ridge Least Trimmed Squares (HRLTS) is efficient than other estimators, using the Standard Error (SE) for real data and Root Mean Squared Error (RMSE) criterion for normal disturbance distribution and different degree of multicollinearity.