Gauss-Newton estimation of parameters for a spatial autoregression model

被引：20

作者：

Bhattacharyya, BB

Khalil, TM

Richardson, GD

机构：

[1] N CAROLINA STATE UNIV,DEPT STAT,RALEIGH,NC 27607

[2] UNIV CENT FLORIDA,DEPT MATH,ORLANDO,FL 32816

来源：

STATISTICS & PROBABILITY LETTERS | 1996年 / 28卷 / 02期

关键词：

martingale central limit theorem; spatial autoregression; unit root estimation;

D O I：

10.1016/0167-7152(95)00114-X

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Estimation of (alpha,beta)' in the doubly geometric model Z(ij) = alpha Z(i-1,j) + beta Z(i,j-1) - alpha beta Z(i-1,j-1) + epsilon(ij) is discussed for the cases (i) alpha = 1, \B\ < 1 and(ii) alpha = beta = 1. In each case, the ''one step Gauss-Newton estimator'' is shown, when properly normalized, to be asymptotically normal.

引用

页码：173 / 179

页数：7

共 12 条

[1] REGRESSION-MODELS WITH SPATIALLY CORRELATED ERRORS [J].