fclosest_ar2.m

来自「空间统计工具箱」· M 代码 · 共 33 行

function [bmax, srds, prhigher,emax, maxlik]=fclosest_ar2(x, y, clo)
%
%[alphamax, loglik, emax, bmax, srds, prhigher]=fclosest_ar2(x, y, closest)
%
%INPUTS:
%
%The matrix x represents n observations on k independent variables.
%
%The vector y represents n observations on the dependent variable.
%
%closest represents either a list of closest neighbor indices such as from fclosest_nn2 
%or a sparse adjacency matrix (s1 when loading smats.mat from fneighbors2).
%
%OUTPUT:
%
%alphamax is a scalar measuring spatial dependence 
%
%maxlik is a scalar giving the maximum of the profile log-likelihood
%
%emax is a n by 1 vector of closest neighbor residuals
%
%bmax is a k by 1 vector of regression parameter estimates
%
%srds is a k by 1 vector of the signed root deviances associated with bmax
%
%prhigher is the probability in repeated sampling of observing a higher srd
%
%NOTES:
%
%Closest neighbor dependence was studied in:
%
%Pace, R. Kelley, and Dongya Zou, 
%揅losed-Form Maximum Likelihood Estimates of Nearest Neighbor Spatial Dependence,