Documentation of cov5

Global Index (short | long) | Local contents | Local Index (short | long)

Function Synopsis

[xy,ccuse,avl] = cov5(x,y)

Help text

 For the 2x2 matrix, only returns a single value...

 function [xy,ccuse,avl] = cov5(x,y)

COV2    Jim's "normalise by N" Covariance matrix of time series that include
        NaNs.
       ^^^^^^^^^^^^^^
       COV2(X), if X is a vector, returns the variance.  For matrices,
       where each row is an observation, and each column a variable,
       COV2(X) is the covariance matrix.  DIAG(COV2(X)) is a vector of
       variances for each column, and SQRT(DIAG(COV2(X))) is a vector
       of standard deviations.   COV2(X,Y) is COV([X Y]).

       See also COV, CORRCOEF, CORRCOEF2, STD.
       J. Little 5-5-86
       Revised 6-9-88 LS
       Copyright (c) 1984-93 by The MathWorks, Inc.

 ***** REVISED BY JR, 27 Apr 1994 - normalises by n, not n-1
 ***** REVISED by ID, 5 Oct 1995 - skips NaNs.
 XY is the covariance matrix and AVL is a row vector with the percentage of
 observations available in each column.

Cross-Reference Information

Listing of function cov5

function [xy,ccuse,avl] = cov5(x,y)

[m,n] = size(x);
if nargin > 1
    [my,ny] = size(y);
    if m ~= my | n ~= ny
        error('X and Y must be the same size.');
    end
    x = [x(:) y(:)];
elseif min(size(x)) == 1
    x = x(:);
end
[m, n] = size(x);
x = x - ones(m,1) * mean2(x);
if m == 1
    xy = 0;
else
    for i=1:n,
       for j=i:n,
          use = find(~isnan(x(:,i))&~isnan(x(:,j)));
          avl(i,j) = length(use);
          avl(j,i) = avl(i,j);
          if avl(i,j) <= 1,
             xy(i,j) = NaN;
          else
             xy(i,j) = x(use,i)'*x(use,j)/avl(i,j);
             xy(j,i) = xy(i,j);
          end
       end
    end
end

% avl is fraction used to determine covariance
avl = (avl/m)*100;
% ccuse is the values used for the correlation coefficient
if n==2
ccuse=find(~isnan(x(:,1))&~isnan(x(:,2)));
end

[mm,nn]=size(xy);
if mm==2 
  xy=xy(1,2);
end