Comparison of the efficiency of some Non-Parametric Methods in estimating Multivariate Linear Regression Models

doi:10.33899/iqjoss.2013.081234

	Comparison of the efficiency of some Non-Parametric Methods in estimating Multivariate Linear Regression Models
IRAQI JOURNAL OF STATISTICAL SCIENCES
Article 23, Volume 13, Issue 3, 2013, Pages 175-197 PDF (0 K)
DOI: 10.33899/iqjoss.2013.081234
Abstract
Regression is one of the statistical methods and widely used in various fields of science in terms of defining clearly the relationship between the variables in the form of an equation to estimate the parameters and the strength and direction of this relationship , also , shows the estimate and predict response. Some nonparametric methods are used for estimating the model parameters of the multivariate linear regression when explanatory variables are ranked , methods and comparisons among them is done using the sum of mean squares error , Data are generated using simulation. The results showed that the method of least squares was better way at increase the sample size in the distributions of Normal and Exponential,where in the Uniform distribution was not affected by not rank and the size of the sample ,but affected variance in the distribution Gamma, the method of Linear ranks was the best in the distribution of the Gamma, but influenced by variance, the method of median weighted was good when the sample size and rank lower only in the normal distribution.

Statistics Article View: 399 PDF Download: 240