1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
|
// Scilab ( http://www.scilab.org/ ) - This file is part of Scilab
// Copyright (C) 2001 - INRIA - Carlos Klimann
//
// This file must be used under the terms of the CeCILL.
// This source file is licensed as described in the file COPYING, which
// you should have received as part of this distribution. The terms
// are also available at
// http://www.cecill.info/licences/Licence_CeCILL_V2.1-en.txt
//
function [v]=mvvacov(x)
//
//This function computes v, the matrix of
//variance-covariance of the "tableau" x (x is a numerical
//matrix nxp) who gives the values of p variables for n
//individuals: the (i,j) coefficient of v is
//v(i,j)=E(xi-xibar)(xj-xjbar), where E is the first
//moment of a variable, xi is the i-th variable and xibar
//the mean of the xi variable.
//
//References: Saporta, Gilbert, Probabilites, Analyse des
//Donnees et Statistique, Editions Technip, Paris, 1990.
//Mardia, K.V., Kent, J.T. & Bibby, J.M., Multivariate
//Analysis, Academic Press, 1979.
//
//
warnobsolete("cov", "6.0.0");
[lhs,rhs]=argn(0)
if rhs<>1 then error(msprintf(gettext("%s: Wrong number of input argument: %d expected.\n"),"mvvacov",1)), end
if x==[] then s=%nan, return, end
[n p]=size(x);
fact=1/n
v = fact * ((x'*x) - fact * ( (ones(1,n)*x)' * (ones(1,n)*x)))
endfunction
|
