Reshape array of cells with different column sizes into matrix

Hi. I need to reshape a T1 = [1x5] cell array, where each cell is a [5x1] cell array of numbers into a S1 = [m-by-n] matrix of numbers. The problem is the initial cell array T1 may have different number of rows in each cell - meaning I don't have a square shaped array to use " cell2mat " function.

Example:

T1=[1x5] cell array
T1{1}   T1{2}   T1{3}   T1{4}   T1{5}
 1       2       3       4       5
 6       7       8       9      10
11      12      13      14      15
16      17      18      19      20
21      22      23      24

I need to add all of the columns together:

T2=
 1   2   3   4   5
 6   7   8   9  10
11  12  13  14  15
16  17  18  19  20
21  22  23  24

And then reshape the array into [m-by-n] array (always square shaped), in this example S1=[3x8]:

S1=
 1  2  3  4  5  6  7  8
 9 10 11 12 13 14 15 16
17 18 19 20 21 22 23 24

I thought that maybe I could reshape T1 into [x-by-1] array and then reshape that array in to S1=[m-by-n]. But what should i do in order to get results like this:

T3=
1
2
3
4
...
24

And not like

1
6
11
16
21
2
...
24

Any help would be appreciated.

[Stephen23]

@Ardo Laaneots: is the shorter vector always the last one in the cell array?

Answer 1

[Jos (10584)]

Here is a way:

T = {[1:5:21]',[2:5:22]',[3:5:23]',[4:5:24]',[5:5:20]'}
m=3
n=8
[M, tf] = padcat(T{:})
M = M.' 
out = reshape(M(tf.'),n,m).' % swap m and n, and then transpose

PADCAT concatenates unequal sized vectors by padding shorter ones with NaNs. This function can be downloaded from the File Exchange: https://uk.mathworks.com/matlabcentral/fileexchange/22909-padcat-varargin-

Answer 2

[KL]

One approach is to make all elements equal in size by padding 0s or nans and then use cell2mat,

m = max(cellfun(@numel,T1));
for k = 1:numel(T1)
    n = numel(T1{k});
    if n<m
        T1{k} = [T1{k}; nan(m-n,1)];
    end
end
T2 = cell2mat(T1)'