Experimental study to compare Robust proposed methods for estimating nonparametric function of cluster data with other methods

Abstract

The statistical analysis aims to assess the impact of the explanatory variables on the response variables during period because the parametric models are subject to many restrictions and borrowings have been resorting to nonparametric models, the nonparametric function was estimated to nonparametric regression design for cluster data that are linked with the same cluster that represents response variables and explanatory variables across different time periods, respectively. It should be noted that the data cluster is similar to longitudinal data in terms of reliability of the data with the same cluster (sector of the longitudinal data) on each other. This paper addressed to estimate the unknown nonparametric data cluster function by using Robust proposed methods to mimic the statistical methods but they are not affected by other influential values, where the performance of classical methods is weak in presence of these values, and compare it with other methods by using criteria standard MSE and MAE, and some of the functions within the simulation experience, which achieved less from the rest of the other function methods by using Huber Function to assess the nonparametric function of the cluster data.Key words: nonparametric regression, cluster data, correlation matrices, Huber Function, MSE, MAE.