How do I replace/convert all zeros in 3rd-dimension to NaN

Hi All,

I have a 5-D matrix: 5x16x20x3x15, where 5x16 is the data set; 20 is the number of trials (per experiment); 3 is the number of experiments (per participant); and 15 is the number of participants. If I had 1 particant that had a dataset (for only one exp) that was exclusively all zeros i.e. 5x16x20x2x2 was entirely zeros; would there be a way to convert the zeros to NaN, so that my mean and std calculations are not affected?

Cheers.

Answer 1

[David Hill]

yourMatrix(:,:,affectedTrials,affectedExperiment,affectedParticipant)=nan;

If all trials of the affected experiment are zeros, then:

yourMatrix(:,:,:,2,2)=nan;

[NA]

Thanks. Is there a way of incorporating this into an if statement, so that I don't have to manually do this for each experiment per participant. I came up with the following but it's not changing the 0 to NaN:

...
    for trl = 1:nTrial
    
    %%load trial data%%
    
    DataCon(:,:,trl,nExp,part) = data(ind1:ind2, ch)*1000; 
    
        if sum(data(:,:),[1 2])==0
           idx(trl)=NaN;
        end
    
    DataCon = DataCon(:,:,idx,nExp,part); 
    
    end 
...

[David Hill]

If your matrix is not too large, this will work fine.

for k=1:size(yourMatrix,3)
  for n=1:size(yourMatrix,4)
    for m=1:size(yourMatrix,5)
      if nnz(yourMatrix(:,:,k,n,m))==0
        yourMatrix(:,:,k,n,m)=nan;
      end
    end
  end
end

[NA]

thanks mate