How to sort cell array of tables by size and concatenate vertically

A cell array consists of individual tables of different sizes. It is to be sorted by size (along row) and then tables (along row) are to be concatenated vertically.

(Note: Sort command with cellfun is not working as reported in earlier versions).

Answer 1

[BhaTTa]

Hey @Mukund, Please refer to the below code that meets your requirement, please take it as a reference and modify it accordingly:

function concatenatedTable = sortAndConcatenateTables(cellArrayOfTables)
% sortAndConcatenateTables Sorts a cell array of tables by row count 
% and concatenates them vertically.
%
% Input:
%   cellArrayOfTables - A cell array where each cell contains a table.
%
% Output:
%   concatenatedTable - A single table resulting from vertical
%                       concatenation of the sorted tables.

% --- Input Validation (Optional but good practice) ---
if ~iscell(cellArrayOfTables) || isempty(cellArrayOfTables)
    error('Input must be a non-empty cell array.');
end
if ~all(cellfun(@istable, cellArrayOfTables))
    error('All elements in the cell array must be tables.');
end

% --- Get the number of rows for each table ---
% We'll use a loop since cellfun with 'size' directly on tables can be tricky
% or less readable for specific dimension extraction.
numRows = zeros(length(cellArrayOfTables), 1);
for i = 1:length(cellArrayOfTables)
    if ~isempty(cellArrayOfTables{i})
        numRows(i) = height(cellArrayOfTables{i}); % 'height' is preferred for tables
    else
        numRows(i) = 0; % Handle empty tables if necessary
    end
end

% --- Sort the tables based on the number of rows ---
[~, sortOrder] = sort(numRows);
sortedCellArray = cellArrayOfTables(sortOrder);

% --- Vertically concatenate the sorted tables ---
% Check if there's anything to concatenate
if isempty(sortedCellArray)
    concatenatedTable = table; % Return an empty table
    return;
end

% Ensure all tables have the same variable names for concatenation
% This is crucial for vertcat. If variable names differ, vertcat will error.
% We'll assume the first table's variable names are the standard.
% More robust error handling or name harmonization could be added here.
if ~isempty(sortedCellArray) && ~isempty(sortedCellArray{1})
    referenceVarNames = sortedCellArray{1}.Properties.VariableNames;
    for i = 2:length(sortedCellArray)
        if ~isempty(sortedCellArray{i}) && ...
           ~isequal(sortedCellArray{i}.Properties.VariableNames, referenceVarNames)
            warning('Table %d has different variable names. Concatenation might fail or produce unexpected results.', i);
            % Optional: Add code here to rename variables or skip the table
        end
    end
end

% Perform the concatenation
concatenatedTable = vertcat(sortedCellArray{:});

% --- Display the result (Optional) ---
disp('Concatenated Table:');
disp(concatenatedTable);

end

% --- Example Usage: ---
% Create some sample tables of different sizes
table1 = table(rand(3,1), rand(3,1), 'VariableNames', {'A', 'B'});
table2 = table(rand(1,1), rand(1,1), 'VariableNames', {'A', 'B'});
table3 = table(rand(5,1), rand(5,1), 'VariableNames', {'A', 'B'});
table4 = table(); % Empty table
table5 = table(rand(2,1), rand(2,1), 'VariableNames', {'A', 'B'});

% Create a cell array of these tables
myTables = {table1, table2, table3, table4, table5};

% Call the function
finalTable = sortAndConcatenateTables(myTables);