I expect this code does what you want. The order in which the array is processed matters, so you need to make sure that this satisfies your edge cases as well.
data=[...
2 6 5 1 3 3
3 8 2 3 5 6
6 3 1 4 2 6
4 6 6 4 3 4];
for col=1:size(data,2)
[~,uniquerows]=uniquetol(data(:,col),2*eps,'DataScale',1);
for row=find(~ismember(1:size(data,1),uniquerows))
%loop through all duplicates (skip if none exist)
%(keeping the first occurence of the value)
forbidden_values=data(:,col);
%add 0 to the list to account for prior iterations
forbidden_values=[forbidden_values;0]; %#ok<AGROW>
candidate_values=data(row,:);
ind=find(~ismember(candidate_values,forbidden_values));
if isempty(ind)
replace_val=0;
else
%pick first option if there are multiple
replace_val=candidate_values(ind(1));
end
data(row,col)=replace_val;
end
end
clc,disp(data)
