How to calculate the mean for specific rows in a cell using cellfun

A cell A{1}=a1 A{2}=a2 ... A{5}=a5

a1=[1 1 1 1 ; 2 2 2 2; 3 3 3 3; 4 4 4 4;]

a2 =

   [ 2     2     2     2;
     3     3     3     3;
     4     4     4     4;
     5     5     5     5;]
...

I want to get the two mean values for the first two rows and the last two rows seperately. Can I realize this using the following function instead of a loop? cellfun(@nanmean,A, 'UniformOutput',false);

Answer 1

[Azzi Abdelmalek]

A={rand(4);rand(4);rand(4);rand(4);rand(4);rand(4)}
n=numel(A)
c=arrayfun(@(x) mean(mean(A{x})),[1 2 n-1 n],'un',0)

[Skirt Zhang]

Hi Azzi,

Thanks for your reply. But i don't have the A{x} with the row nr. = col nr. . The size for each A{x} is 10*4000.

So I run your code and can't get the idea result that I want, which should be Mean(first 5 rows in A{x}) and Mean(last 5 rows in A{x}).

Can you help me further about this?

Thanks

I don't store in 3 D due to the memory problem.

[Azzi Abdelmalek]

Skirt, in your question you said 2 first rows and two last rows, From your post, A is a cell array of n elements, each element is a matrix. Can you explain clearly what is the expected result (look at my example)

[Azzi Abdelmalek]

If you wat first 5 and last 5, just adapt the above code

A={rand(4);rand(4);rand(4);rand(4);rand(4);rand(4)}
n=numel(A)
nr=5
c=arrayfun(@(x) mean(mean(A{x})),[1:nr n-nr+1:n],'un',0)

Answer 2

[Matt J]

Can I realize this using the following function instead of a loop? cellfun(@nanmean,A, 'UniformOutput',false);

You can, but there will be no advantage in performance over the for loop.

Incidentally, if all the A{i} are the same size, as in your example, it's puzzling that you wouldn't maintain them as a 3D array A(:,:,i). Then you could just do

nanmean(A([1,end],:,:),2)

Answer 3

[José-Luis]

It is not really clear how you want to split your data, so I added two variables, rowsFromStart and rowsFromEnd so you can specify it yourself. It is also not very clear if you want one mean for each row or the mean of all rows. Finally, a loop would probably be faster.

 A={rand(10,100);rand(10,100);rand(10,100);rand(10,100);rand(10,100);rand(10,100)}
rowsFromStart = 5;
rowsFromEnd = 5;
c=cellfun(@(x) [mean(mean(x(1:rowsFromStart-1,:),2));mean(mean(x(end:end-rowsFromEnd+1,:),2))],A,'un',0)