Comparative Study Of Manly Transformation And Yeo-Johnson Transformation In Quantile Regression Analysis
Nwakuya Maureen Tobechukwu
Department of Mathematics and Statistics, University of Port Harcourt, Rivers State, Nigeria.
Biu Emmanuel O
Department of Mathematics and Statistics, University of Port Harcourt, Rivers State, Nigeria.
Keywords: Yeo-Johnson transformation, Manly Transformation, , Chamberlain and Buchinsky two stage method, quantile Regression
Abstract
This paper tries to implement Yeo-Johnson and Manly transformations in a quantile regression analysis while using the Chamberlain and Buchinsky two stage estimator to estimate the transformation parameter for both methods. A comparison was done between the two methods. The comparison was based on the bias of the estimates, AIC and MSE. The results of this analysis revealed that at all quantiles considered (0.05, 0.25, 0.5, 0.75 and 0.95), the Yeo-Johnson transformed quantile regression performed better than the Manly transformed quantile regression based on the AIC and MSE. The bias of the estimates also indicated that the Yeo-Johnson transformed Quantile regression had better results except at the 25th quantile where the Manly method had a better result.