How can I concatenating 2 table

left = table(['A0';'A1';'A2'],['B0';'B1';'B2'],... 'VariableNames',{'A' 'B'},... 'RowNames',{'K0' 'K1' 'K2'}); right = table(['C0';'C2';'C3'],['D0';'D2';'D3'],... 'VariableNames',{'C' 'D'},... 'RowNames',{'K0' 'K2' 'K3'});

Allname = union(left.Properties.RowNames, right.Properties.RowNames); result = zeros(length(Allname),size(left,2)+size(right,2)) * NaN; table_result= array2table(result); table_result.Properties.VariableNames = [left.Properties.VariableNames right.Properties.VariableNames]; table_result.Properties.RowNames=Allname; index_A =ismember(Allname, left.Properties.RowNames); index_B =ismember(Allname, right.Properties.RowNames); table_result(index_A,left.Properties.VariableNames) = A <- ERROR

I want to solve the problem. Give them a hand.

Answer 1

[Ameer Hamza]

The column entries of the table must belong to the same class, therefore you can't use nan since it is double, whereas other elements are char arrays. In the following code, I have replaced it with char array 'nan'. Other than that, the following script will work as required.

left = table(['A0';'A1';'A2'],['B0';'B1';'B2'],...
    'VariableNames',{'A' 'B'},...
    'RowNames',{'K0' 'K1' 'K2'});
right = table(['C0';'C2';'C3'],['D0';'D2';'D3'],...
    'VariableNames',{'C' 'D'},...
    'RowNames',{'K0' 'K2' 'K3'});

Allname = union(left.Properties.RowNames, right.Properties.RowNames);
result = repmat({'nan'}, length(Allname), size(left,2)+size(right,2)); %zeros(length(Allname),size(left,2)+size(right,2)) * NaN;
table_result= array2table(result);
table_result.Properties.VariableNames = [left.Properties.VariableNames right.Properties.VariableNames];
table_result.Properties.RowNames=Allname;
table_result(left.Properties.RowNames,left.Properties.VariableNames) = table2cell(left);
table_result(right.Properties.RowNames,right.Properties.VariableNames) = table2cell(right);

Answer 2

[Peter Perkins]

You want outerjoin.

Answer 3

[Sterling Baird]

outerjoin() works in the case you described (no common variable names between the tables), but when that's not the case, it will produce different behavior than what you seem to be going for. For example, with these two tables as inputs:

nrows = 3;
%define table columns
doubles = rand(nrows,1);
chars = repelem('a',nrows,1);
cells = repelem({rand(10)},nrows,1);

%make two tables
tbl1 = table(doubles,chars)
tbl2 = table(chars,cells)

>>

tbl1 =

  3×2 table

    doubles     chars
    ________    _____

     0.75906      a  
    0.017267      a  
     0.49258      a  


tbl2 =

  3×2 table

    chars        cells     
    _____    ______________

      a      {10×10 double}
      a      {10×10 double}
      a      {10×10 double}

outerjoin(tbl1,tbl2) will produce the following:

  9×4 table

    doubles     chars_tbl1    chars_tbl2        cells     
    ________    __________    __________    ______________

     0.75906        a             a         {10×10 double}
     0.75906        a             a         {10×10 double}
     0.75906        a             a         {10×10 double}
    0.017267        a             a         {10×10 double}
    0.017267        a             a         {10×10 double}
    0.017267        a             a         {10×10 double}
     0.49258        a             a         {10×10 double}
     0.49258        a             a         {10×10 double}
     0.49258        a             a         {10×10 double}

As a simple modification to outerjoin, you can assign a list of temporary IDs that you remove at the end. You just need to makes sure that 'ids' (or whatever you choose) doesn't overlap with a variable in one of the tables.

t1names = tbl1.Properties.VariableNames;
t2names = tbl2.Properties.VariableNames;
sharednames = intersect(t1names,t2names);
tbl1.ids = (1:3).';
tbl2.ids = (4:6).';
tbloutjoin = outerjoin(tbl1,tbl2,'Key',['ids',sharednames],'MergeKeys',true);
removevars(tbloutjoin,'ids')

>>

    doubles    chars        cells     
    _______    _____    ______________

    0.24666      a      { 0×0  double}
    0.08822      a      { 0×0  double}
    0.47188      a      { 0×0  double}
        NaN      a      {10×10 double}
        NaN      a      {10×10 double}
        NaN      a      {10×10 double}

You may also consider using FEX: tblvertcat, which builds on outerjoin() following a similiar method as shown and supports 2+ tables: tblvertcat(tbl1,tbl2)

    doubles     chars        cells     
    ________    _____    ______________

     0.75906      a      { 0×0  double}
    0.017267      a      { 0×0  double}
     0.49258      a      { 0×0  double}
         NaN      a      {10×10 double}
         NaN      a      {10×10 double}
         NaN      a      {10×10 double}