การเปรียบเทียบตัวประมาณการถดถอยเมื่อมีพหุสัมพันธ์และ/หรือมีค่าผิดปกติ / ปัทมวดี นันทนาเนตร์ = A comparison of regression estimators when multicollinearity and/or outliers are present / Pattamawadee Nantananet
The objective of this research is to compare the regression estimators of parameters in a multiple linear regression model when multicollinearity and/or outliers are present. The regression estimators are the Least Squares Estimators (LS), Least Absolute Value Estimators (LAV), Ridge Estimators (RID), Ridge Least Absolute Value Estimators (RLAV), and Weighted Ridge Estimators (WRID), The comparison was done under the following conditions. The distributions of random errors are Normal Distribution and Contaminated Normal Distribution. The size of the outliers of dependent variable are small, medium, and large level according to the proportion of the contaminations of random errors are 5%, 8%, 10% and 15%. The level of multicollinearity are 0.1, 0.3, 0.5, 0.7, 0.9, 0.95 and 0.99 , and the sample size are 20, 30, 35, 40, 50, and 60. The data of this experiment were generated through the Monte Carlo Simulation Technique. The experiment was repeated 1,000 time under each condition to calculate the square root of the mean squares error (RMSE) of each regression estimator.
