Get all unique combinations of 2 columns from a table

I have a table that has 4 variables, similar to this:

    Item1    Item2    Item3    Item4
    _____    _____    _____    _____

      1        5        7       10  
      

What is the best way to produce the delta of each combination of columns? Ex. Item1 vs Item2, Item1 vs Item3, etc.

Answer 1

[Stephen23]

V = {'Item1','Item2','Item3','Item4'};
T = table(1, 5, 7 ,10, 'VariableNames',V)

T = 1×4 table
    Item1    Item2    Item3    Item4
    _____    _____    _____    _____

      1        5        7       10  

X = nchoosek(1:numel(V),2);
D = diff(T{:,V}(X),1,2);
Z = cell2table([V(X),num2cell(D)])

Z = 6×3 table
      Var1         Var2       Var3
    _________    _________    ____

    {'Item1'}    {'Item2'}     4  
    {'Item1'}    {'Item3'}     6  
    {'Item1'}    {'Item4'}     9  
    {'Item2'}    {'Item3'}     2  
    {'Item2'}    {'Item4'}     5  
    {'Item3'}    {'Item4'}     3  

Answer 2

[Cris LaPierre]

I would look into using the perms function, along with the unique and diff functions.

[Austin B]

I couldn't find a good way to succinctly use perms or unique to get my desired output, but I may just not be well versed enough in MATLAB yet to do so. I was able to come up with a solution that in my opinion solved this rather cleanly:

t = table(1, 5, 7 ,10, 'VariableNames', {'Item1' 'Item2' 'Item3' 'Item4'});
tDelta = table();
varNameStrings = string(t.Properties.VariableNames);
for v1 = varNameStrings
    for v2 = varNameStrings(find(varNameStrings==v1):end)
        comp = v1+" vs "+v2;
        if v1 ~= v2
            tDelta.(comp) = diff([t.(v1) t.(v2)]);
        end
    end
end
disp(t)

    Item1    Item2    Item3    Item4
    _____    _____    _____    _____

      1        5        7       10  

disp(tDelta)

    Item1 vs Item2    Item1 vs Item3    Item1 vs Item4    Item2 vs Item3    Item2 vs Item4    Item3 vs Item4
    ______________    ______________    ______________    ______________    ______________    ______________

          4                 6                 9                 2                 5                 3       

[Cris LaPierre]

I was probably trying too hard to make use of the variablenames. Stephen's approach is simpler by using column numbers instead.

T = table(1, 5, 7 ,10, 'VariableNames', {'Item1' 'Item2' 'Item3' 'Item4'});
C = nchoosek(T.Properties.VariableNames,2)

C = 6×2 cell array
    {'Item1'}    {'Item2'}
    {'Item1'}    {'Item3'}
    {'Item1'}    {'Item4'}
    {'Item2'}    {'Item3'}
    {'Item2'}    {'Item4'}
    {'Item3'}    {'Item4'}

delta = diff([T{1,{C{:,1}}};T{1,{C{:,2}}}])

delta = 1×6

4 6 9 2 5 3

[Austin B]

His and yours are both excellent and elegant, thank you!