How do I combine uneven matrices into a single matrix?

Question

dt1 dt2 dt3 dt
1   3   6   10
2   4   1   5 
3   6   5   3
4   7   4   1
5   1   2   4
6   2   8
7   8
8
9
10

I have the above data which I want to combine in one single matrix (10 x 4). The maximum number of rows is 10. I created a zeros matrix. However, I have a problem since the data doesn't have the same dimensions. How it is possible to get the output as below? Should I sort the data and replace the missing values with 0?

1   1   1   1  
2   2   2   0  
3   3   0   3   
4   4   4   4   
5   0   5   5  
6   6   6   0   
7   7   0   0
8   8   8   0
9   0   0   0
10  0   0   10

possible duplicate of [Accumulating cells of different lengths into a square matrix in Matlab](http://stackoverflow.com/questions/3054437/accumulating-cells-of-different-lengths-into-a-square-matrix-in-matlab). All you need to do is group your variables into a cell array first. — Jonas, Jul 04 '10 at 12:39
I have another problem, the number of variables are incremental e.g. in a loop, how I can group it? — Jessy, Jul 04 '10 at 13:40
Oops, looks like this is similar, but no full duplicate of the question I linked to. Sorry. — Jonas, Jul 04 '10 at 13:44

score 2 · Answer 1 · edited May 23 '17 at 10:24

2

Here is the modified version of @gnovice's answer to a previous, similar question

%# group the variables. If you would be generating them in a loop, you could use the loop
%# to group them, i.e. have something like
%# for i=1:n
%#    dtCell{i} = "function that generates dt_i"
%# end

dtCell = {dt1,dt2,dt3,dt};
nCells = length(dtCell);
maxVal = max(cellfun(@max,dtCell)); %# this way, I don't have to know vector orientation

%# you could replace the loop with calls to cellfun. 
%# While this may make you feel more Matlab-ish, it wouldn't be
%# faster, or more readable
out = zeros(maxVal,nCells);
for iCell = 1:nCells
   idx = dtCell{iCell}; %# this assignment is just for clarity 
   out(idx,iCell) = idx;
end

edited May 23 '17 at 10:24

Community

1
1

answered Jul 04 '10 at 13:54

Jonas

74,690
10
137
177

1

Nice, but you need to change the computation of the maximum value to something like `maxVal = max(vertcat(dtCell{:}));` to get the maximum value across *all* cells. – gnovice Jul 04 '10 at 16:07
@gnovice: Thanks for the heads-up! Fixed now. – Jonas Jul 04 '10 at 17:02

score 0 · Answer 2 · answered Apr 16 '15 at 21:09

Let's define example data:

dt1 = randi(10,10,1)-1;
dt2 = randi(10,7,1)-1;
dt3 = randi(10,6,1)-1;
dt4 = randi(10,5,1)-1; %// example data. Column vectors.

One approach is to create a mask with bsxfun and then fill values at the positions indicated by the mask:

dt = {dt1, dt2, dt3, dt4}; %// collect data into a cell array
n = cellfun(@numel, dt); %// length of each vector
mask = bsxfun(@le, (1:max(n)).', n); %// create mask
result = zeros(size(mask)); %// initiallize result with zeros
result(mask) = vertcat(dt{:}); %// fill in values.

Example result (with random data):

result =
     2     3     2     7
     3     9     8     6
     4     7     9     0
     4     3     5     7
     2     1     1     1
     1     7     7     0
     8     0     0     0
     1     0     0     0
     7     0     0     0
     8     0     0     0

How do I combine uneven matrices into a single matrix?

2 Answers2

Linked