How do I query 200 cells containing datasets of unequal lengths using one data having complete data?

I have this code that overwrites nan values (stored in refom) with data from different cells (A). A has 200 cells each containing datasets

The problem is when I use this for loop below to loop through the data in the cells it does overwrite the nan values as I expect

Nevertheless, in rows where there is no data I expect it to leave the NANs but what happens is that it rather put some random

values in positions where there is no data.

On the other hand when I try to do it out of the for loop using just two datasets it works.

What is missing please?

rf = {};
refom = nan(size(A{1})); %A{1} has full dataset hence using its size to create NANs
for i = 1:length(A)
    r_new = A{i}(:,:);
    [~,rows] = ismember(r_new(:,3),A{1}(:,3)); %compare col 3 of the two data
    refom(rows,:) = r_new; % overwrite NaNs and keep NANs for rows without data 
    rf = [rf; refom]; %append to cell dustbin rf
  
end

Answer 1

[Jan]

The data are not random, but the values of the former iterations. You do not reset the contents of refom to NaN in each iteration.

refom = nan(size(A{1}));
match = A{1}(:, 3);   % [EDITED] 3 column only
for i = 1:numel(A)
  r_new = A{i};
  [~,rows] = ismember(r_new(:,3), match);
  refom(rows, :) = r_new;
  rf = [rf; refom];
  refom(:) = NaN;
end

[Stephen Tete]

Im using column 3 to query the whole dataset please edit

[~,rows] = ismember(r_new(:,3), match(:,3));

The first one was very helpful, thank you alot