Create table from data

Hi

I'm trying to extract some data (A1-4) and place it into a table. I get the error message "Error using vertcat - Dimensions of arrays being concatenated are not consistent." Can you advise what this is and how I can fix it?

The overall script the below code is from is more complex. The purpose of it is to extract data from a large number of statistical tests from multiple subjects and place the data in one table. When I run the script on one file (one subject and one test) I get the same error message. When I remove T1-4 from within the [ ] below and run it on one file then the table comes out fine. However, if I do that with all my files in a loop then the table comes out empty. For the one file I tested it on; A1-3 are 1x94 double and A4 is 1x94 cell.

 T1 = [T1; A1]';
 T2 = [T2; A2]';
 T3 = [T3; A3]';
 T4 = [T4; A4]';

T = table(T4, T1, T2, T3)

[Adam Danz]

One of the lines is trying to combine data with a different number of columns. If this is expected, you can pad the matrix with fewer columns (padarray). If this is unexpected, perhaps the problem occurs when reading in the data.

[DavidL88]

Hi Adam

Thank you for your reply. I had thought the same but it's not clear to me how there could be a different number of columns. When I run the script, A1-4 all have the same number of columns. When I remove T from within the [] ie T1 = [A1] then the table T comes out fine so there's no problem combining it then (I seem to need the T within [] for looping as this was required in other scripts I ran).

[Adam Danz]

> ...it's not clear to me how there could be a different number of columns

How/why there are a different number of columns is a different question from wether there is a different number of columns or not. You should test whatever line is causing the error by directly looking at the number of columns using,

% (example for T1)
size(T1,2)
size(A1,2)  

> When I run the script, A1-4 all have the same number of columns.

Do they have the same number of columns as T1? Is T1 empty at the start of the series of concatenations? BTW, why not simplify the vertical concatenation using,

T = [T1;A1:A2;A3;A4];
% or, if T1 is initially empty, 
T = [A1:A2;A3;A4];

> When I remove T from within the [] ie T1 = [A1] then the table T comes out fine

That suggests that T1 does not have the same number of columns as A1. It would be helpful to know what line is causing the error.

To help further, I need to know the size of all T1, A1..A4 and, if they are tables, it might be helpful to see their first few rows (using head()).

[Walter Roberson]

T = [A1;A2;A3;A4];

I suspect you intended.

[DavidL88]

Hi Adam

I have the T variables as T1 = []; at the start of the script. It says the line causing the error is T1 = [T1; A1]'; I ran size as you suggested. It comes out different as below which would suggest this is the issue (it would be simpler to just run without T variables but it doesn't seem to work then when I loop it for a group of test files).

size(T1,2) = 1
size(A1,2) = 94

I ran T = [A1;A2;A3;A4]; and got the below message.

Error using vertcat
Dimensions of arrays being concatenated are not consistent. Consider converting input arrays to the same type before
concatenating.

Error in working_file (line 33)
T = [A1;A2;A3;A4];

All the A's for this one test are 1x94 double so 1 row and 94 columns. All the T's are blank.

For clarity, I copy the entire script below. Removing the T variables and just using T = table(A4, A1, A2, A3) after 'end' gives what I'm looking for when sFiles = one test. But when sFiles = all tests I get back a blank for all the A variables (A = []) and for the table (T = 0×4 empty table).

% Process: Select files using search query
sFiles = subject/test 

 T1 = [];
 T2 = [];
 T3 = [];
 T4 = [];

for iFile = 1:length(sFiles)
 DataMat = in_bst_data(sFiles(iFile).FileName);
 
 tvalue = DataMat.tmap
 pvalue = DataMat.pmap
 time = DataMat.Time
 
 col = (find(DataMat.pmap(1,:)<0.05));
  
 [X Y]=size(col);
 
 A1 = (tvalue(:,col));
 A2 = (pvalue(:,col));
 A3 = (time(:,col));
 A4 = (repmat({DataMat.Comment},1,Y'));
  
 T1 = [T1;A1]';
 T2 = [T2;A2]';
 T3 = [T3;A3]';
 T4 = [T4;A4]';
 
end

T = table(T4, T1, T2, T3)

[Adam Danz]

> I have the T variables as T1 = []; at the start of the script.... size(T1,2) = 1

There you go. T1 doesn't equal [] when you are concatenating it.

T = []; 
size(T)

ans = 1×2

0 0

●

[Adam Danz]

This would be a lot easier if you attach a mat file with the variables you're concatenating.

[DavidL88]

I solved it for one test by deleting all T = [] and using T1-4 = (cell(X,Y)'); within the loop so the T variables are the same size as the A variables.

I placed this after the line with [X Y]=size(col);

 T1 = (cell(X,Y)');
 T2 = (cell(X,Y)');
 T3 = (cell(X,Y)');
 T4 = (cell(X,Y)');

When I set it sFiles to equal all test files the table comes out empty. The script is set so that only p values = 0.05 and their corresponding t values and latencies are returned. The last file in the loop has no p values <0.05. I'm not sure why the other values are not captured in the table T though. I can't upload the mat file because it's private data (from research). Thank you for your help.

[Adam Danz]

It's getting more difficult to follow what's-what. It also seem like you're making the mistake of trying anything to fix the problem rather than understanding the problem.

From what I understand you have 4 arrays, A1 A2 A3 and A4 and you'd like to vertically concatenate those arrays which requires them to have the same number of columns but you're getting an error when you try to do this.

If the A# arrays should have the same number of columns but they do not, the solution is figuring why they don't have the same number of columns.

If the A# arrays might not have the same number of columns and you want to vertically concatenate them, you need to either trim the extra columns or pad the narrower arrays to match the max number of columns. Are you using Matlab's debug mode? If not, this would be a good time to learn by setting a break point just before the concatenation and looking at the values of the arrays (see demo).

[DavidL88]

Hi Adam,

I solved it. Thanks for your help and the tip about debugging (I had not used it before). You are right that I wanted to concatenate all the A arrays side by side into one table. I had two issues which were linked but I wasn't sure how to fix. Concatenate the A variables with T variables and do this in a for loop. All the A arrays were the same size but usually >1 so couldn't seem to concatentate them with T variables which were set to = 1. I tried to set the T variables to same size as As within each for loop (or to remove the T variables and simplify the script) but the goal I was seeking, a table with the data from each iteration would be blank (it was bing written over with each iteration). I simply replaced the T1-4 = []; with T1-4 = cell(0,0);, placed before the for loop, and this seems to have fixed the issue.

[Adam Danz]

High-five!

I didn't know a cell could be concatenated to a table.

% cell-table concatentation
[num2cell(rand(1,3)); array2table(rand(3,3))]

ans = 4x3 table
     Var1        Var2       Var3  
    _______    ________    _______

    0.76622    0.054515    0.67136
     0.9878     0.71826    0.13674
    0.93978     0.66781    0.75822
    0.64564     0.35228    0.40694

By the way, you could have set T to T=[] or T=table(), too.

[table(); array2table(rand(3,3))]

ans = 3x3 table
     Var1       Var2       Var3  
    _______    _______    _______

    0.17549    0.64745     0.8014
    0.18883    0.42064    0.92881
    0.53757    0.93268    0.49124

But I still don't know why T=[] didn't work for you,

[[]; array2table(rand(3,3))]

ans = 3x3 table
      Var1       Var2       Var3  
    ________    _______    _______

     0.56871    0.88387    0.58633
    0.080156    0.79495    0.56984
     0.19062    0.60072    0.69567

[DavidL88]

Thank you!

I wanted to accept your answer but can't see the option next to your comment?

[Adam Danz]

I copied the comment I think you're referencing to the answers section. Thanks for keeping the forum tidy!

Answer 1

[Adam Danz]

Summary of comments under the question,

Instead of initializing T as an empty cell, use an empty table.

T = table(); 
A1 = array2table(rand(3,3));

C = [T,A1]

C = 3x3 table
     Var1       Var2       Var3  
    _______    _______    _______

    0.73391    0.42024    0.11362
    0.81381    0.81976    0.94401
    0.23762    0.68696    0.64335