Comparison of some methods of estimating non-parameter regression model using simulation

Abstract

In this paper, the abilities of the non-parametric regression function model were studied
by taking the Local Linear Nonparametric Estimator and the Local Quadratic
Nonparametric Estimator using the Leave - One-Out Cross - Validation criterion. The
robust Smoothing Spline Estimator (RSS) was developed using the RCV, and simulation
had to be used as an optimal tool to study sample sizes and different random error
distributions, including normal and contaminated data with abnormal or contaminated
values and varying levels. Different and for different models, to compare Av
DOCUMENTS assessment studied the approved and knowledge of the best methods,
have demonstrated results destined preference primaries slides fortified (Robust
Smoothing Spline) (RSS) using standard hippocampus (RCV) by receiving a lower value
to the standard average relative rate of absolute values of errors AMAPE)).