proportion of gender having health issue

I have a table that has Gender vector (1=Boy, 2=Girl) and Healtissue (based on score). I would like to know the proportion of girls only have a health issue if their score more than 17?

[dpb]

Look at findgroups, groupsummary with the Subject "Split-Apply-Combine" workflow under splitapply

Many examples...

[Doaa Alamoudi]

Thank you

Answer 1

[Image Analyst]

Assume t is your table and you have a column for gender, a column with HealthScore (true or false), and a column for their score (continously-valued number). Then try

% Find rows with girls who have a score more than 17.
rows1 = (t.Gender == 2) & (t.Score > 17) 
% Now find girls with a score more than 17 who ALSO have a health issue (may not be 100% - may only be part of them).
rows2 = (t.Gender == 2) & (t.Score > 17) & t.HealthIssue  % Assuming this is a logical column.
proportion = rows2 / rows1

[Doaa Alamoudi]

Thanks for your help

I have created the following

score=sdq;% obtaining scores from sdq vector
Gender= (sex == 1);% To get male student only  
count=0;

for K=1:length(score)
    O=1:length(Gender);
    
   MentalHealthProblem = score > 17 & Gender ; % count male whos score >17 

        count=count+1;

     
end

[Image Analyst]

Uh, okay. But you said you wanted girls. And it was not clear how the health problem was defined. I thought it was independently determined, like some percentage of those with scores above 17 would have a problem. But it looks like you're saying that if the score is more than 17, they definitely, 100% have a health problem. If that's it, then that's fine. Anyway, this code doesn't look good. Your count will always equal length(score) so what's the point? Plus MentalHealthProblem will always be a logical vector, assuming sdq was a vector, not a scalar, which seems reasonable since you're doing a loop for every element of it. Do you even need a for loop if you simply want

MentalHealthProblem = sum(score > 17 & Gender); % Count males whose score > 17 

[Doaa Alamoudi]

Great that give good out put

now if you want to compare the result of boys who are under the smae condition with the girls who have same problem how can I do that?

[Image Analyst]

I'd do

maleRows = (sex == 1);  % To get male students only.
femaleRows = (sex == 2); % To get female students only.

[Doaa Alamoudi]

I rewrite based on your suggestion and I think it looks bettern now. However I have problem in creating figure showing the result

score=sdq;
BoyGender= (sex == 1);
GirlGender=(sex==2);
BoysMentalHealthProblem = sum(score > 17 & BoyGender)
GirlsMentalHealthProblem = sum(score > 17 & GirlGender)

Answer 2

[dpb]

Some other things to explore...

hScore=randi(100,100,1)/4;                                          % An artificial dataset...
sex=(rand(size(hScore))>0.5)+1;                                     % 50:50 roughly
tMH=table(sex,hScore,'VariableNames',{'Gender','HealthIssue'});     % make a table
tMH.Gender=categorical(tMH.Gender,[1,2],{'Boy','Girl'});            % turn into categorical instead
tMH.AtRisk=categorical(tMH.HealthIssue>17);                         % compute the risk factor
heatmap(tMH,'Sex','AtRisk')                                         % one way to look at results...

Results will vary for a random sample, but for the particular dataset generated here:

>> groupsummary(tMH,'Gender')
ans =
  2×2 table
    Gender    GroupCount
    ______    __________
     Boy          51    
     Girl         49    
>> groupsummary(tMH,'AtRisk')
ans =
  2×2 table
    AtRisk    GroupCount
    ______    __________
    false         68    
    true          32    
>> groupsummary(tMH,{'Gender','AtRisk'})
ans =
  4×3 table
    Gender    AtRisk    GroupCount
    ______    ______    __________
     Boy      false         37    
     Boy      true          14    
     Girl     false         31    
     Girl     true          18    
>> 

You can compute percentages from the GroupCounts depending upon whether want by Gender or overall...the above produces

[Doaa Alamoudi]

That was Great answer

Thanks for your help