How do I sort the rows of a matrix by a certain column, given the ordering specified in a separate vector?

How do I sort the rows of a matrix by a certain column, given the ordering specified in a separate vector and without a for loop?

For example we have A and B:

A =[ 3 1 5]

B =[ 5 55

       3    33

       1    11

       3   333

       5   555

       1   111]

and wish to reshape B by the first column using the order specified in A to give C:

C =[3 33

      3   333

      1    11

      1   111

      5    55

      5   555]

Answer 1

[MathWorks Support Team]

There are likely many ways to do this without a for loop. One way is to cast the first column of matrix to a Categorical data type. We can think of the first column as a discrete set of categories rather than a numeric datatype. Then by specifying these categories as ordinals, we can rank them, and use this information to sort. See the below steps to do this using the example:

% The example data:
A=[3 1 5]; %the indexing vector
B=[5 55;3 33;1 11;3 333;5 555;1 111]; %the matrix to be sorted

% Step 1 convert the first row of B to a categorical datatype and save to a
% temporary variable.
B_Cat = categorical(B(:,1),A,'Ordinal',true);
% Note: We use A to define the category names. By also
% specifying it as an Ordinal, the order of categories specified in A
% determines the ranking used when sorting or comparing.

% Step 2: sort the temporary variable, save the indices for rearranging B.
[~,I] = sort(B_Cat);

% Step 3: Rearrange B by using the indices from the previous step to create
% the desired C matrix
C = B(I,:)

The output is:

C =

       3    33
       3   333
       1    11
       1   111
       5    55
       5   555