Find rows based on set of values/codes

Hello community,

I have a data set that is 48x14 (called data for this example). Each row has a set of codes (possible codes are 0 1 13 14 15 16 31 42 53 54). a row might look like this: [1 0 0 13 0 0 31 0 0 0 42 0 0 53].

The code is part of a for loop but I am strugglig with one specific part: I want to find rows that have a specific series of codes while ignoing the 0 codes. I thought the following code would work to find rows with the codes 1 13 31 42 and 53 in it:

for i = 1:length(data(:,1))

if all(data(i,find(data(i,:) == [1 13 31 42 53])))

.......

.......

I get the following error:

Arrays have incompatible sizes for this operation.

Any help would be much appreciated!

Answer 1

[Star Strider]

I am not certain what you want, however the ismember function might be a better option than find for this, especially since everything seems to be integers, so no floating-point approximation problems will be present (requiring ismembertol and likely more coding) —

data =  [1 0 0 13 0 0 31 0 0 0 42 0 0 53];
q = ismember(data, [1 13 31 42 53])

q = 1×14 logical array
   1   0   0   1   0   0   1   0   0   0   1   0   0   1

v = data(q)

v = 1×5

1 13 31 42 53

Then determine what you want to do with either ‘q’ (using the nnz function could give the number of matches) or ‘v’ (the content of the matches in the vector) here.

.

[Millie Johnston]

I am not looking for the sequnce per say, I am looking for the rows in which those numbers appear together (in any order). Does that make more sense?

[Star Strider]

Yes, it does, and it actually makes the problem easier.

If you want to test for all of them being present, then the if, elseif, else test woulld be:

if nnz(q) == 5

If they are not all present, then the nnz result will be less than 5.

If the vector changes its length, then the code becomes (a bit more robustly):

testvct = [1 13 31 42 53];

q = ismember(data, testvct);

if nnz(q) == numel(testvct)

These could certainly be combined into fewer lines. I kept them separate here to illustrate the concept.

.

[Millie Johnston]

Excellent! Thank you very much!

[Star Strider]

As always, my pleasure!

Answer 2

[David Hill]

for i = 1:size(data,2) 
   if all(ismember(data(i,:),[0 1 13 31 42 53]))%include 0 

[Image Analyst]

This answer doesn't care about the order of the desired sequence of codes you are looking for or if there are any duplicated numbers in the row.

My answer below requires that the sequence match exactly (not be scrambled or in arbitrary order like [13, 53, 42, 31, 1]).

We're not sure which way you want it, but now you have it both ways so you can choose.

[Millie Johnston]

I am not looking for the sequnce per say, I am looking for the rows in which those numbers appear together (in any order). Does that make more sense?

Answer 3

[Image Analyst]

Try this

oneRow =  [1 0 0 13 0 0 31 0 0 0 42 0 0 53];
oneRow(oneRow == 0) = [] % Remove zeros

oneRow = 1×5

1 13 31 42 53

In addition you should be using isequal

% Define the specific sequence you are looking for.
desiredCodeSequence = [1 13 31 42 53]; % Row should be exactly this.
% See if it matches your zero-removed row.
if isequal(oneRow, desiredCodeSequence)
    % Matched
    fprintf('Matched.\n')
else
    % The row is not the desired sequence.
    fprintf('The row is not the desired sequence.\n')
end

Matched.