For loop through cell arrays

I have a 1×4 cell array of

{60×4 double} {60×4 double} {60×4 double} {60×4 double}

and I need to;

-find the values at the 3rd column at the rows have "20" at the first column in each cell

-then eliminate zeros if there is any

-then take average

As an illustration

          10         NaN           0         NaN                     20         NaN         623           1         
          10         NaN           0         NaN                     20         NaN           0         NaN
          20         NaN         310           1                     10         NaN          38           1
          20         NaN           0         NaN                     10         NaN           0         NaN
          20         NaN        1445           1                     10         NaN           0         NaN

the first cell should give the average of 310 and 1445, whereas the second cell should give 623.

I am thinking of a for loop but I couldnt make it work. Thanks in advance!

[Adam Danz]

Could you share your loop so we can help fix it? Otherwise we're doing it for you.

[Ali Yilmaz]

Hi Adam,

Honestly I deleted my version as it was not even close to what I need. I'll be glad if you can share your version.

Thanks and regards

[Adam Danz]

See answer below.

Answer 1

[Adam Danz]

If you want to do this in a loop, there are 4 steps I've outlined below.

"c" is the 1x4 cell array.

column3Means = nan(size(c)); 
for i = 1:numel(c)
    % Find rows where column 1 equals 20
    index1 = c{i}(:,1) == 20; 
    
    % Find rows where column 3 is not 0
    index2 = c{i}(:,3) ~= 0; 
    
    % Combine those indices
    idx = index1 & index2; 
    
    % Use the combined index to compute the mean
    column3Means(i) = mean(c{i}(idx,3),'omitnan');
end

FYI, the loop can be avoided using the cellfun() function.

[madhan ravi]

Wow didn’t know there was a cellarray() function , was it released in 2020a?

[Adam Danz]

oops :D typo. I meant cellfun.

Corrected. Thanks, madhan.

[Ali Yilmaz]

Hi Adam,

Thank you very much for your response.

When I apply your code, column3means turns out to be [NaN,NaN,NaN,NaN]

[Ali Yilmaz]

Here I added the cell array, now I have 1x5 cell array. But the question is the same.

-find the values at the 3rd column at the rows have "20" at the first column in each cell

-then eliminate zeros if there is any

-then take average

[Adam Danz]

Could you share the code you're using to compute the mean?

You probably just need to work in the "nanflag" input to ignore nans in your data.

M = mean(___,nanflag)

[Ali Yilmaz]

my cell array is zeroback

column3Means = nan(size(zeroback)); 
 for i = 1:numel(zeroback)
    % Find rows where column 1 equals 20
    index1 = zeroback{i}(:,1) == 20; 
    
    % Find rows where column 3 is not 0
    index2 = zeroback{i}(:,3) ~= 0; 
 end
    mean(index2, 'omitnan')

Now I have a mean value of 0.25.

But actually I need the rows which have 20 at 1st column whilst the value in their 3rd column is nonzero. However, with this code above I feel like I have the rows that have 20 at their first coulmn and the rows that have a nonzero value at their 3rd column separately.

[Adam Danz]

You're not using index1. You're also taking the mean of the index rather than using the index to extract the values from zeroback.

But please look at every line and look at the values being produced on each line so you can understand what they are doing.

Try this.

column3Means = nan(size(c)); 
for i = 1:numel(c)
    % Find rows where column 1 equals 20
    index1 = c{i}(:,1) == 20; 
    
    % Find rows where column 3 is not 0
    index2 = c{i}(:,3) ~= 0; 
    
    % Combine those indices
    idx = index1 & index2; 
    
    % Use the combined index to compute the mean
    column3Means(i) = mean(c{i}(idx,3),'omitnan')
end

[Ali Yilmaz]

Thank you very much Adam;

that worked perfectly, appreciate your help

[Adam Danz]

Glad it worked out.

The take-home message is: learn/use indexing.

Indexing is a superpower of Matlab. My answer uses logical indexing but there are other types of indexing as well.

https://www.mathworks.com/company/newsletters/articles/matrix-indexing-in-matlab.html

[Ali Yilmaz]

Thank you so much again, I will definitely go throuh the link you've shared.