Matrix Data Results Explanation

I used the following code, and you can see that in the second row of the fourth column, the result is zero and not NaN; it should be NaN as the result.

%
soma_total = nan(3, 4);

data1 = [nan nan 1 2; 1 2 nan nan; 1 2 nan nan];

data2 = [5 5 nan nan; 2 1 1 nan; nan nan 1 1];

% 
result = nansum(cat(3, soma_total, data1, data2), 3);

% 
disp(result);

The result was:

[5 5 1 2;

3 3 1 0;

1 2 1 1]

The result should be:

[5 5 1 2;

3 3 1 NaN;

1 2 1 1]

Does anyone have an idea to solve this?

Answer 1

[Dyuman Joshi]

nansum does not do what you are expecting it to do. Also, using nansum() is not recommended as you can see from its documentation page. Use sum instead -

Here's a workaround -

%
soma_total = nan(3, 4);

data1 = [nan nan 1 2; 1 2 nan nan; 1 2 nan nan];

data2 = [5 5 nan nan; 2 1 1 nan; nan nan 1 1];

%Dimension to operate on
dim = 3;

%defining the array
arr = cat(dim, soma_total, data1, data2);

%Finding indices for all elements which are NaN corresponding to the "dim" dimension
idx = all(isnan(arr),dim)

idx = 3×4 logical array
   0   0   0   0
   0   0   0   1
   0   0   0   0

%If you want to use nansum instead, use this command
%result = nansum(arr,dim)
result = sum(arr,dim,'omitnan')

result = 3×4

5 5 1 2 3 3 1 0 1 2 1 1

%Converting the respective values back to NaN
result(idx) = NaN

result = 3×4

5 5 1 2 3 3 1 NaN 1 2 1 1

[Augusto Gabriel da Costa Pereira]

Thanks, mister.

[Dyuman Joshi]

You are welcome!