How can I remove NaN values from a 7862x6 cell?

Hello, Matlab community:

I have a 7862x6 cell (variable name = A) that has NaN values at the end of some of the columns of data (hundreds of consecutive data points). How can I remove the NaNs so that I can find the true length of each column? I have looked through many current threads, but those solutions have not worked for my purpose. I think the problem is that I need a loop to work through the columns and rows, though I am not sure how to approach it. Any suggestions are appreciated!

- John

[Vrajeshri Patel]

You can't remove them because it would change the size of the matrix.If you just want to find the length of the column (not including nan entries), here's an example:

   k=[5 6 nan; 1 nan 3; 1 2 2];%example matrix
   k(find(~isnan(k)))=1;%replace all non-nan numbers with 1s
   s=nansum(k,1);%sum the values

Answer 1

[dpb]

Presuming can decipher the actual storage you're trying to describe, here's a sample and solution to the number...

>> A=nan(10,3); for i=1:3,A(1:randi(8,1),i)=i;end  % create some dummy data
>> A=num2cell(A)
A = 
    [  1]    [  2]    [  3]
    [  1]    [  2]    [  3]
    [  1]    [  2]    [  3]
    [  1]    [  2]    [  3]
    [  1]    [  2]    [  3]
    [  1]    [NaN]    [  3]
    [  1]    [NaN]    [NaN]
    [  1]    [NaN]    [NaN]
    [NaN]    [NaN]    [NaN]
    [NaN]    [NaN]    [NaN]
>> sum(isfinite(cell2mat(A)))  % number non-NaN each column
ans =
     8     5     6
>>

Since A is a cell array, you can remove those rows on a column-by-column basis if desired but isn't necessary to determine the sizes and may be more convenient if don't 'cuz can convert as-is to an 'ordinary' array which can make referencing and calculations simpler, depending on what else is needed.

[John Harry]

Thank you very much, dpb! Your suggestion did exactly what I need!

[dpb]

Actually, the "real" way would be

>> sum(cellfun(@isfinite,A))
ans =
   8     5     6
>>

No need for the cell2mat which perhaps makes a copy. I've not tested performance...