estimating the nonparametric regression function study monotone nonparametric methods for

Abstract

This research was concerning to study monotone nonparametric methods for estimating the nonparametric regression function (i.e treatment outlier) to achieve a monotone function (increasing or decreasing).So we will use the monotone methods to treatment outlier but after estimate the regression function with use kernel estimator (Nadarya - Watson) these methods are:-1- Mukerjee method takes averages of maximums and minimum of subsets of the data was used to adjust the initial kernel regression estimates and use the researcher special case when .2- Algorithm least square isotonic regression.In the experimental aspect comparison was done of which is the best methods through the simulation procedure using Mote Carlo method using five models.While in the application aspect practical application was done on data represent the measurements for blood pressure patients.In both aspects we use two of the important statistical measures which are Mean square error (MSE) and efficiency. We find through the application that the best method is Mukerjee method for general case as it has minimum Mean square error and maximum efficiency.