A New Efficient Redescending M-Estimator for Robust Fitting of Linear Regression Models in the Presence of Outliers

Table 1

Total mean square error for datasets generated from normal distribution (X & Y∼N (0, 1)).

N	P	OLS	Huber	Hampel	Andrew	Tukey	Proposed
N = 30	P = 2	0.301	0.301	0.301	0.301	0.301	0.301
	P = 3	0.339	0.350	0.342	0.369	0.369	0.342
	P = 6	0.523	0.547	0.550	0.741	0.731	0.527
N = 50	P = 2	0.196	0.204	0.200	0.206	0.012	0.197
	P = 3	0.248	0.256	0.250	0.260	0.260	0.249
	P = 6	0.362	0.381	0.373	0.371	0.411	0.363
N = 100	P = 2	0.138	0.144	0.141	0.144	0.144	0.108
	P = 3	0.175	0.180	0.175	0.180	0.180	0.170
	P = 6	0.242	0.258	0.253	0.261	0.261	0.207