creating dummy variables in a dataset

Hi

I'm very new to Matlab...so thanks for your patience. I'm working with the hospital dataset that comes with Matlab (load hospital in Matlab). I want to create a new vector, say DummyGender by looking at the vector hospital.Sex. The rule is:

for x = 1:100;
    if hospital.Sex=='Male';
        gender{x}=1;
    else
        gender{x}=-1;
    end
end

this works, but I need to go ahead and transpose it, and then convert it to a numerical variable and then finally add it as another column in the dataset. Is there a more efficient way to do the same?

thanks C

[Georgios Tertikas]

did you find a solution? i have the exact same problem

[Chet]

Hi George - haven't had a chance to look at this in a while. Do take a look at Walter's response below...maybe that would be useful to you.

Answer 1

[Walter Roberson]

transpose() is one way of writing the transpose operation.

It appears you are already creating a numeric value. 1 and -1 are numeric. If you mean that the cell array is to be converted to a numeric array, why not store it as a numeric array in the first place?

Efficient way? I would suggest

gender = zeros(100,1);
for x = 1  : 100
   if strcmp( hospital(x).Sex, 'Male')
     gender(x) = 1;
   else
     gender(x) = -1;
   end
end

after which no transpose operation is needed as it will already be a column vector.

Even more efficient would be just

Male_tf = strcmp( {hospital.Sex}, 'Male');
gender = 2 * Male_tf(:) - 1;

[Chet]

Thanks Walter! On your comment " If you mean that the cell array is to be converted to a numeric array, why not store it as a numeric array in the first place?" - the dataset that Matlab provides (load 'hospital') has the data as a string to begin with...and I'm trying to convert that to numeric. The code you provided did not work...and that's because I provided the wrong information (like I mentioned, I'm new to Matlab). Male and Female in the database are nominal variables. I guess that's why my original code worked (I don't think one can work with Strings using '==", the code I had up there worked because 'Male' and 'Female' are nominal variables, not String variables). So I need to start from scratch again. Sorry for wasting your time, and thanks for your help again. Thanks Chet

BTW - the dataset hospital came preloaded on my version of Matlab

http://www.mathworks.com/help/stats/dataset-array-columns.html