Extracting rows from a matrix based on values in columns

Following on from this question

http://au.mathworks.com/matlabcentral/answers/224877-how-to-extract-rows-based-on-column-values-in-a-matrix

I have a large number of conditions lets say i needed 28 different values in the columns to find, how would i do that without finding the indices in 28 different lines, not sure how to do the for loop? :/

Answer 1

[Stephen23]

This is easy using accumarray:

A = [3, 0.1234;
  1, 0.1345;
  1, 0.1456;
  2, 0.1567;
  1, 0.1678;
  1, 0.1789];
C = accumarray(A(:,1),A(:,2),[],@(n){n});

Which places each collection into one cell of cell array C:

>> C
C = 
    [4x1 double]
    [    0.1567]
    [    0.1234]
>> C{1}
ans =
       0.1345
       0.1456
       0.1678
       0.1789

This will automatically create as many cells as needed in one output variable C. You need to use cell array indexing to access the contents of the cells of C. Note that you should not attempt to create 28 separate variables and access them dynamically in a loop: this is very poor programming practice (sadly much loved by beginners, who think it is great) that is slow, buggy, and obfuscated:

http://www.mathworks.com/matlabcentral/answers/196952-how-to-create-new-variables-in-batches-with-strcat

[Maria445]

What if I have more than 1 column? That is, I want A to be a 6-by-4 matrix for example, so to have C{1} = 4-by-3 matrix?

[Stephen23]

@Maria445: make the second input to accumarray the row indices, and change the anonymous function to use those indices to extract the required rows:

>> R = 1:size(A,1);
>> C = accumarray(A(:,1),R.',[],@(r){A(r,:)})