sirf.m

计量工具箱

function [xindex, beta, valp]= sirf(x,y,nbcla)
% PURPOSE: SIR method for a unidimentional output (Sliced Inverse Regression)
%--------------------------------------------------------------
% USAGE: [xindex, beta, valp] = sirf(x,y,nbcla)
% where:   x =  explaining variable (matrix nxp)
%          y =  variable to explain (vector nx1)
%          nbcla = number of classes (integer)
%--------------------------------------------------------------
% OUTPUTS: xindex = indices = x'Beta (matrix nxp)
%          beta = vectors of edr direction (column) (matrix pxp)
%          valp = eigen values (vector px1)
%--------------------------------------------------------------
% written by Yves Aragon & Christine Thomas-Agnan (2002)
% Universit