CODEWITHSUNDEEP
matlabmatrixmultidimensional-arraymeanstandard-error
Aiskya
Aiskya

Reputation: 73

How to calculate the standard error for multiple matrices in MATLAB?

I want to calculate the standard error of multiple matrices concatenated into one:

cd C:\User\Aisk_000\Desktop\A\NC\Subjects\2014A\
A = dlmread('weights.txt');
cd ../3169A
B = dlmread('weights.txt');
cd ../3350A
C = dlmread('weights.txt');
cd ../3607A
D = dlmread('weights.txt');
cd ../4073A
E = dlmread('weights.txt');


%concatenate
x = cat(5,A,B,C,D,E)
y = mean(x,5)       %calculate the average
z = std( x ) / sqrt( length( x ))   %Calculate standard error

The output from z is a 5-D double, same as the one for x. However, I realized that the system calculated the standard error of each matrix instead of the overall concatenated matrix. I know that it has something to do with the number "5", but I'm not sure where to put it in the code for z.

Upvotes: 2

Views: 756

Answers (1)

gnovice
gnovice

Reputation: 125854

The first argument to cat isn't the number of matrices you are concatenating, it's the dimension along which you want to concatenate them. Since your matrices are all 2-D 16-by-16 matrices, it makes sense to concatenate them along the third dimension, stacking them into a 3-D matrix.

x = cat(3, A, B, C, D, E);

You can then use mean and std, specifying 3 as the dimension along which to perform the calculation:

y = mean(x, 3);                      % Mean along dimension 3
z = std(x, 0, 3)./sqrt(size(x, 3));  % Standard error along dimension 3

Upvotes: 1

Related Questions