Removing nan values inside cells of a cell array and keeping the og structure

Hi,

I have looked into

https://se.mathworks.com/matlabcentral/answers/105236-remove-nans-in-matrices-within-cell-array

and

https://se.mathworks.com/matlabcentral/answers/517798-removing-nan-values-inside-cells-of-a-cell-array

However, as for my problem. I have a [1Xn] cell, where within it one of these is a [kXv] cell. It is easy to extract each of the cells in the cell array and remove all NAN, but I want to keep the origninal structure. Therefore I am asking if it is possible to have a simple line of code that doesent split up all of the cells in the cell array so I dont have to paste them toegether at a later stage?

[Ameer Hamza]

The solution depends on how you data is structured? Can you share your cell array?

[Erik From]

Certain rows contain NAN, and I want to remove these rows, while still keeping this structure for a later part of my code.

Answer 1

[Image Analyst]

Why would extracting those and removing nans change the original structure of the cell array? Why can't you just do this:

for k = 1 : numel(raw)
	ca = raw{k}; % Get k'th cell of the raw cell array.
	[rows, columns] = size(ca);
	% ca is a cell array of character strings.
	for col = 1 : columns
		for row = 1 : rows
			contents = ca{row, col};
			% You cannot remove a cell from a cell array 
			% but you can set it's contents to null instead of NaN.
			if isnan(contents)
				contents = [];
			end
		end
	end
end

[Saud Alfalasi]

Hi IA.

Please may you look at this question:

https://www.mathworks.com/matlabcentral/answers/668383-recombine-cell-arrays-to-produce-a-new-image?s_tid=mlc_ans_email_ques

[Erik From]

Some of the commands i tested put the values that were not NAN into one row. I employed your accepted answer in https://se.mathworks.com/matlabcentral/answers/318627-how-to-find-nan-values-in-a-cell-array, with a simple change to only work for the first row. This is the result:

Would giving the row a null number change the value of size, some entire columns are NAN and I only want to "replace"/remove ones that coincide with the first column. It seems that the code you recommended replaces every single NAN? Further:

to insert ca into raw, would

raw{k}=ca

suffice?

[Image Analyst]

Yes. Since you forgot to include your variable, I had to do it without testing. I forgot that you have to stick contents back into ca, and then ca back into raw

ca{row, col} = contents; % Put repaired contents back into the cell.
raw{k} = ca; % Put the cell ca back into the cell array raw.

Attach your raw if you need more help.

save('answers.mat', 'raw');

Then use the paperclip icon to attach answers.mat.

[Erik From]

Here is also a simple screenshot.

Since I at a later stage want to pair certain dates according to an equation and then remove them, according to something like this:

Salv={'2020-11-16 11:02:30'};

Constant=1;

for b=1:length(Salv)

for k=1:length(raw)

x = raw{1,k}(:,4);

rowNumber = strfind(x,Salv(Constant));

raw{1,k}(rowNumber,:) = [];

end

Constant=Constant+1;

end

It does not like that I have both cells and num together for the strfind.