How do I use a for loop to index a categorical vector?

Here is my code:

hydrocephalus_1 = ismember(DRG, {'31', '32', '33'});
idx = find(hydrocephalus_1 == 1);

icdpcs_values = zeros(length(idx), 1);
for i = 1:length(idx)
    icdpcs_values(i) = i10_pr1(idx(i));
    i = i+1;
end

[GC, GR] = groupcounts(icdpcs_values)

i10_pr1 is a categorical vector. I am trying to see the values in i10_pr1 that exist at the locations of 31, 32, and 33 in DRG. The result is coming out as different values than are in i10_pr1. The values that are coming out are just numbers when they should be a mix of numbers and letters. Is this issue arising because i10_pr1 is a categorical vector? Can someone help debug this so that it runs properly? Thanks for the help!!

[dpb]

Be MUCH easier to do if you would attach the data...the positions in a categorical array are indexable same as any other vector, but the values of idx are the locations relative to positions in the DRG vector which may not be correlated w/ i10_pri 1:1. We can't tell that; but that's what your result seems to say is so...

[caroline lesniak]

I cannot attached the data because it is confidential, but the DRG vector is correlated 1:1 with i10_pr1. They are different columns from the same data table.

[dpb]

Throw some random values in in place of the real thing then. But somehow need to be able to have a test case that reproduces your issue.

Answer 1

[dpb]

If the two indeed correlate, the result should be as expected -- ergo, can only presume they don't correlate as you think...example that simulates your code:

DRG=randi(40,20,1);                                 % a sample for DRG
i10_pr1=categorical(cellstr(char('A'+[0:19]')));    % a categorical variable to carry along with DRG
>> tCL=table(DRG,i10_pr1)                           % create a table from both variables
tCL =
  20×2 table
    DRG    i10_pr1
    ___    _______
    38        A   
    20        B   
    20        C   
    14        D   
    37        E   
    15        F   
     5        G   
    32        H   
    16        I   
    10        J   
    17        K   
     4        L   
     6        M   
    38        N   
    39        O   
    24        P   
     3        Q   
    10        R   
    15        S   
    33        T   
>> tCL(ismember(tCL.DRG,[31,32,33]),:)    % the outputs for desired/looked-for rows
ans =
  2×2 table
    DRG    i10_pr1
    ___    _______
    32        H   
    33        T   
>> tCL(ismember(tCL.DRG,[31,32,33]),'i10_pr1')  % return the categorical variable only
ans =
  2×1 table
    i10_pr1
    _______
       H   
       T   
>> 

I used unique categories above to make identifiable as being the specific rows in question.

As noted above, to produce something that replicates what you think is an error will require having a test/sample case that illustrates the problem; no way to know how to reproduce your symptoms without being able to see what is really going on.

[caroline lesniak]

length(DRG)

ans =

     3117413

>> length(i10_pr1)

ans =

     3117413

Both are the same length. I did a test case with random numbers and letters for i10_pr1 and had i10_pr1 and DRG be the same length and my code failed again.

I just tried using your new code from above, but it is not working for me. I am getting this error:

Error using categorical/ismember (line 69)
Both inputs to ISMEMBER must be categorical arrays, or one of the inputs
may be a scalar string or cell array of character vectors.

Error in code_sortDRG_new (line 3)
tCL(ismember(tCL.DRG,[31,32,33]),:)

[caroline lesniak]

I figured it out thank you for the help!! This was very helpful. I just had to convert [31, 32, 33] to a categorical array.

[dpb]

OK. Knew it had to be something straightforward but not being able to see your terminal from here makes debugging somewhat more difficult...