How can I transpose a dataset or table?

How can I transpose a dataset or table (make the rows and columns switch) in MATLAB R2013b?

For example, I have the following dataset:

 >> X = dataset({[1;2], 'a'}, {[100;200], 'b'}, 'ObsNames', {'c','d'})

 X =

          a    b  
     c    1    100
     d    2    200

and I would like to transpose the dataset:

 >> Xt = X'

 Xt = 

          c      d  
     a      1      2
     b    100    200

Similarly, if I have the same data stored in a table:

 >> X = array2table([1 100; 2 200],'VariableNames',{'a','b'},'RowNames',{'c','d'})

 X = 

          *a*     *b* 
          _    ___

     *c*    1    100
     *d*    2    200

and I would like to transpose the table:

 >> Xt = X'

 Xt = 

           *c*      *d* 
          ___    ___

     *a*      1      2
     *b*    100    200

How can I do this?

Answer 1

[MathWorks Support Team]

The ability to transpose a dataset or table using the transpose operator (') is not available in MATLAB R2013b, however this is possible using a combination of other commands.

For datasets, you can use a combination of 'dataset2cell' and 'cell2dataset' to convert the dataset into a cell array, transpose the cell array, then translate back into a dataset:

 >> Xc = dataset2cell(X)

 Xc = 

     'ObsNames'    'a'    'b'  
     'c'           [1]    [100]
     'd'           [2]    [200]

 >> Xt = cell2dataset(Xc','ReadObsNames',true)

 Xt = 

          c      d  
     a      1      2
     b    100    200

For tables, you can use a combination of 'table2cell' and 'cell2table':

 >> Xc = table2cell(X)

 Xc = 

     [1]    [100]
     [2]    [200]

 >> Xt = cell2table(Xc','RowNames',X.Properties.VariableNames,'VariableNames',X.Properties.RowNames)

 Xt = 

           c      d 
          ___    ___

     a      1      2
     b    100    200

 As of MATLAB R2018a, you can also use the ROW2VARS function: 

 <https://www.mathworks.com/help/matlab/ref/rows2vars.html>

[Peter Perkins]

Converting to a cell array and back is probably not a good idea unless the table is fairly small. If the goal is to transpose the data in the table, it's a safe bet that all the data are the same type. Use table2array and array2table instead.

Beginning in R2018a, the best way to to this is rows2vars, as that method deals with things like variable names.

dataset is a class is the Statistics & Machine Learning Toolbox. If possible, it's a good idea to use tables instead.

[Peter Perkins]

"Beginning in R2018a, the best way to to this is rows2vars, as that method deals with things like variable names."