Correlation with two matrices

Hi everyone,

I would like to find the correlation about two matrices the same dimensions. I have two matrices with 64 channles and I need to find the correlation between the first value of first matrix and the first, second, third.... until sixty-four value of second matrix and viceversa.

How can I do this?

Thanks a lot!

Answer 1

[David Hill]

You could just manually do it yourself.

A=randi(100,10);B=randi(100,10);
M=mean(A);N=mean(B);
r=zeros(size(A,2),size(B,2));
for m=1:size(A,2)
    for n=1:size(B,2)
        r(m,n)=sum((A(:,m)-M(m)).*(B(:,n)-N(n)))/sqrt(sum((A(:,m)-M(m)).^2.*(B(:,n)-N(n)).^2));
    end
end
r

r = 10×10

-0.7325 -1.5279 0.5441 -1.1353 1.3671 -1.9351 -1.4018 0.8595 0.3189 0.8992 0.0661 0.1537 0.7952 0.6901 -0.5238 0.0602 -1.4856 1.2302 -0.9858 0.1995 -1.1532 -2.2351 1.1573 1.5280 -0.2541 -0.0752 0.8163 0.5013 -0.5230 -0.5775 0.1300 -0.4017 -0.7430 -0.1586 0.9928 -1.7321 0.4399 -0.2708 -0.7359 -0.3293 -0.8045 -1.8755 1.9652 2.4100 -0.1757 -0.6482 -0.6868 1.0841 0.0492 -1.6882 -0.1862 0.4648 -0.0476 -0.4808 0.2425 -0.5745 -0.6061 1.5960 0.0200 0.6626 -0.8411 -0.0155 0.0680 -0.9309 1.0928 0.8641 -0.5821 0.6830 0.5620 -0.2508 -0.4532 -0.6661 1.2019 1.5538 -0.9566 0.6813 1.0478 -0.1999 1.4598 -1.1943 -0.0501 -0.1554 -0.2710 -0.9457 -0.5791 0.9445 0.1837 0.7055 -0.0926 1.7323 -2.1825 -0.8110 2.0911 1.4211 0.6435 -1.3528 -0.0415 0.1698 2.3680 -1.8121

Answer 2

[Bora Eryilmaz]

You can compute the pairwise correlation between columns of matrices as follows:

% Data matrics with 64 channels (columns)
A = rand(10,64);
B = rand(10,64);

% Vector of pairwise correlation values
C = corr(A,B); % All pairs of columns
C = diag(C) % Matching pairs of columns

C = 64×1

0.1045 0.2102 -0.1445 -0.0380 0.3045 -0.2592 0.2869 -0.5510 0.1050 0.0790

[Felicia DE CAPUA]

Thanks, but if I have this cycle for:

% Inputs

L = channels_1;

R = channels_2;

% Preallocate memory for output

correlations = cell(size(L));

% Loop over each cell

for ncell = 1:size(L,1)

% Define the number of correlation coefficients in this cell

ncorr = size(R{ncell},2);

% Preallocate the correlation vector

correlations{ncell} = zeros(1,ncorr);

% Loop over each pair of columns

for ncoeff = 1:ncorr

% Calculate the correlation coefficient

correlations{ncell}(ncoeff) = corr(L{ncell}(:,ncoeff),R{ncell}(:,ncoeff));

end

end

How can I insert the diag?

[Bora Eryilmaz]

Your code seems unnecessarily complicated. If you already have the matrices L and R, the correlations variable need not be a cell array. Something like this should work:

% Data matrics with 64 channels (columns)
L = rand(100,64);
R = rand(100,64);

% Vector of pairwise correlation values
C = corr(L,R); % All pairs of columns
correlations = diag(C) % Matching pairs of columns

corerlations = 64×1

0.1035 -0.0462 -0.0224 -0.0031 -0.0396 -0.1127 0.0125 -0.0144 0.0199 -0.1041

The second loop that goes over the columns of L and R matrices is not needed since the corr() command already handles that for you, given the whole matrices. Unless of course if your L and R "matrices" have a more complicated structure, like being cell arrays, etc.

[Felicia DE CAPUA]

Thanks a lot!