How to average over a cell-array of arrays?

Question

I have a cell array c of equal-sized arrays, i.e. size(c{n}) = [ m l ... ] for any n. How can I get the mean values (averaging over the cell array index n) for all array elements in one sweep? I thought about using cell2mat and mean but the former does not add another dimension but changes l to l*n. And looping manually of course takes like forever...

Answer 1

I found an easy way to find the mean values within a Cell array on the following link: http://www.gomatlab.de/cellfun-t25114.html

May x be the cell. Then:

var_mean = cellfun(@mean, x, 'UniformOutput', false); %columnwise mean value


var_mean = cellfun(@(in) mean(in(:)), x); %% mean value of the total "subcell"

Answer 2

If all of your arrays are the same size, it makes more sense to store them in a matrix rather than a cell array. That makes it easier to perform operations across them, like taking the mean. You can convert your data to a matrix using the functions NDIMS and CAT:

dim = ndims(c{1});          %# Get the number of dimensions for your arrays
M = cat(dim+1,c{:});        %# Convert to a (dim+1)-dimensional matrix
meanArray = mean(M,dim+1);  %# Get the mean across arrays

Answer 3

This just loops through the cell and means the array down until it is a singleton. It doesn't take that long, this is 40 million floats being meaned, takes 1 second.

function n = big_mean
tic
c = cell(1000);

for ii = 1:length(c)
    c{ii} = rand(8,7,6,5,4,3,2);
end

n = all_cells(c);
toc
end

function n = all_cells(c)

n = zeros(length(c),1);
for ii = 1:length(c)
    n(ii) = cell_mean(c{ii});
end

n = mean(n);
end

function n = cell_mean(n)

while length(size(n))~=2
    n = mean(n);
end

end

Elapsed time is 1.042459 seconds.

ans =

    0.4999

Answer 4

If you have a Higher version of matlab, It can be done by 'cellfun' function. This can treat the cells with unequal size array.

C = {1:10, [2; 4; 6], []};
Cmeans = cellfun(@mean, C)
Cmeans =
    5.5000    4.0000       NaN

Reference: https://groups.google.com/forum/?fromgroups=#!topic/comp.soft-sys.matlab/S_hyHxy11f0

Answer 5

You're on the right track. Use CELL2MAT to convert your cell array to a numerical array and then RESHAPE to construct a three dimensional matrix. You can then calculate the mean using the MEAN function with the dimension argument:

>> c = {[1 2 3; 4 5 6] [7 8 9; 12 13 14]}

c = 

    [2x3 double]    [2x3 double]

>> mean(reshape(cell2mat(c), [2, 3, 2]), 3)

ans =

     4     5     6
     8     9    10

Answer 6

thanks for your other comments, but sometimes, it is hard to rearrange the data or change the way they are saved. For those of you who have this issue, here is the solution, Enjoy.

a=0;
MyCellAddFun=@(Input) a*eye(size(Input))+Input;
temp=arrayfun(@(ind) MyCellAddFun(CellData{ind}),1:length(CellData),'uniformoutput',false);
answer=temp{end}