Extract and code gender string as number using if loop

I have data for 200 participants stored in a 1x1 structure with 3 fields (time data, gender and age). I'm trying to include lines of script that will extract the gender of each participant and code their gender as 0 (non-binary), 1 (female), or 2 (male). I'm also assigning an ID and extracting age, code so far is below:

for i = 1:size(files);

id = str2num(files(i).name(7:size(files(i).name,2)-4));

tempfile = load(files(i).name);

age = tempfile.data.age;

end

I don't have experience in using an if-loop on strings and am struggling to find the best command to use. I have tried the below, as this is what I would do if extracting and coding numbers, but obviously this doesn't work for these strings as they are not matched in size. I've also played around with STRCMP and STRNCMP but with no luck.

if tempfile.data.gender == 'non-binary'

gender = 0;

elseif tempfile.data.gender == 'female'

gender = 1;

elseif tempfile.data.gender == 'male'

gender = 2;

A little help would be much appreciated!

Answer 1

[Star Strider]

I do not udnerstand the reason strcmp or strncmp would not work with your structure data. We may have to see ‘tempfile.data.gender’ in order to understand the problem.

Meanwhile, consider strcmpi to do case-insensitive comparisons.

[Jen]

It isn't that strcmp or strncmp didn't work I just couldn't figure out how to format the command properly. I tried to use online guidance and the help function but I couldn't wrap my head around it.

[Bob Thompson]

The basic usage of strcmp is to specify where you are looking and then what you are looking for. For your case it would be something similar to the following.

if strcmp(tempfile.data.gender,'non-binary')
    gender = 0;
end

I would agree that strcmpi is better if there are people making individual inputs (it will make accounting for different variations of the entry easier), and I assume that the entry format is similar.

[Star Strider]

Try this:

if strcmpi(tempfile.data.gender,'non-binary')
        gender = 0;
    elseif strcmpi(tempfile.data.gender, 'female')
        gender = 1;
    elseif strcmpi(tempfile.data.gender, 'male')
        gender = 2;
end

I cannot test this with your data, so I am labelilng it UNTESTED CODE. It should work, however I cannot guarantee that.

[Jen]

Ah yes, that works perfectly and much more concise than my method. Thank you!

[Star Strider]

As always, my pleasure!

Answer 2

[Bob Thompson]

Just to clarify, you have 'gender' as two separate variables, once as a variable called 'gender' which will have a numeric assignment, and once as a field in a nested structure?

strcmp and == will ultimately have the same result, but you are going to run into the size issue on both of them. I work around this by having a second condition to be met for size.

  if length(tempfile.data.gender) > 6 & tempfile.data.gender == 'non-binary'
        gender = 0;
    elseif length(tempfile.data.gender) > 4 & tempfile.data.gender == 'female'
        gender = 1;
    elseif length(tempfile.data.gender) == 4 & tempfile.data.gender == 'male'
        gender = 2;

In a similar manner, if those are guaranteed to be the only possible responses, you can just use the size comparison to assign the value.

[Jen]

(reposted as comment)

Thanks for posting! Yes you're right in that there are two variables called gender at play here - one nested in my data structure, the other in my output as a coded version of the first.

I ran your suggestion but still get the error 'Matrix dimensions must agree' though I can see why this workaround should work, as in theory matlab would only be comparing strings of 6 or more for the first loop etc.

In the mean time though I have come up with the following solution which gives me the output I was hoping for:

STR = tempfile.data.gender;

for p = "n"

if startsWith(STR,p) == 1

gender = 0;

end

end

for p = "f"

if startsWith(STR,p) == 1

gender = 1;

end

end

for p = "m"

if startsWith(STR,p) == 1

gender = 2;

end

end

I'm looking to make this work with the most concise code possible so any thoughts on this are very welcome! I know there's always many ways to achieve the same goal in Matlab, and my code may be a little clunky though it gets the job done.