How to find differences between 2 large tables that have a mix of string and numbers?

Hello! I have 2 very large tables that have the same number of rows and columns (20441x30), and I would like to know exactly which cells are different. I have tried 'Compare' in the Home tab, but it only tells me in the difference summary 'added', not which values/variables were added). I have also tried setdiff and ismember (error below), but it has a problem with one of my variables (maybe because it has NaNs too in the table).

Error using cell/union>cellunionR2012a (line 242)

Input A of class double and input B of class cell must be cell arrays of character vectors, unless one is a character

vector.

Do you know how I can compare tables that have a mix of string and numbers?

Thank you!

[Dyuman Joshi]

"(maybe because it has NaNs too in the table)"

FYI - MATLAB treats NaNs (as well as NaTs, <undefined> and <missing>) values as not equal to each other, nor equal to themselves as well.

If you want to compare NaNs (and others) and output true, use isequaln

Answer 1

[Image Analyst]

Have you tried iterating over every element of both tables using isequal to determine whether the locations are the same? Something like (untested)

[rows, columns] = size(table1);
matches = false(rows, columns); % Initialize to no matches (no equality).
for col = 1 : columns
    for row = 1 : rows
        % See if this location is equal.  
        % Set equality map to true if they're equal.
        if isequal(table1{row, col}, table2{row, col})
            matches(row, col) = true;
        end
    end
end

Alternatively you can compare the numeric columns using table2array and subtracting.

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

[Tanika Bawa]

This worked well, thank you very much =)

Answer 2

[Peter Perkins]

I thnk looping over every row in the tables will not be very performant. I would think this will do better:

matches = false(size(table1));
for var = 1 : width(table1)
    matches(:,var) = (table1.(var)==table2.(var)) ...
                     || (ismissing(table1.(var)) && ismissing(table2.(var)))
end

That assumes the variables support ismissing; you might need a try/catch in general.

[Voss]

Should use & and |

[Tanika Bawa]

thank you! =D

[Peter Perkins]

Yes, Voss is correct, & and |, not && and ||.