Accession Number : ADA185695
Title : Strong Consistency and Exponential Rate of the 'Minimum L1-Norm' Estimates in Linear Regression Models.
Descriptive Note : Technical rept.,
Corporate Author : PITTSBURGH UNIV PA CENTER FOR MULTIVARIATE ANALYSIS
Personal Author(s) : Wu, Yuehua
PDF Url : ADA185695
Report Date : Jun 1987
Pagination or Media Count : 27
Abstract : This document considers a linear regression model, where (x sub i) is a sequence of experimental points, i. e., known p-vectors, (e sub i) is a sequence of independent random errors, with med(e sub i) =0,i= 1,2....Define the minimum L1 -norm estimate of (alpha, beta)', by (alpha, beta)', to be chosen such that under quite general conditions on (x sub i) and (e sub i), the strong consistency of the minimum L1 -norm estimate is established. Further, under an additional condition on (x sub i), it is also proved that for any given epsilon 0, there exist constant C O not depending on n.
Descriptors : *LINEAR REGRESSION ANALYSIS, *MATHEMATICAL MODELS, *ESTIMATES, ERRORS, EXPONENTIAL FUNCTIONS, RATES, NORMAL DISTRIBUTION, CONSISTENCY
Subject Categories : Statistics and Probability
Distribution Statement : APPROVED FOR PUBLIC RELEASE