CROSS-VALIDATION AND MEDIAN CRITERION

In this paper a new method of selecting the smoothing parameter in non parametric regression called median cross validation (MCV) is suggeste d. This method is applied to choose the number of nearest neighbors us ed in estimating the regression function by local sample medians. Unif orm strong consistency is obtained under reasonable conditions. MCV is effective in dealing with outliers in the data. Simulation results ar e given to demonstrate its superiority over other methods.