Correct me if I’m wrong, but for this dataset I thought that treated patients have a label of 1 and untreated patients have a label of 0.
Why then is the expected output of this function 0.6?
To get this output, untreated samples have to be filtered with x[2] == 1 (the treated patient label) and treated samples have to be filtered with x[2] == 0 (the untreated label).
Shouldn’t we be filtering the untreated samples with x[2] == 0 and treated samples with x[2] == 1?
If I do it this way, which seems like the correct way, then the output of the test function is actually 0.4 instead of 0.6.
Please advise
1 Like
What you are thinking is right. And it is okay if your expected output has a little change. your assignment will have the marks.
1 Like
Hi,
I have the same question : I get to an output of 0.6 instead of the expected 0.4.
As a result the unit test is not passing (because Wrong output, makes sense), and the evaluation of Logistic Regression further down is 0.4588 instead of the expected 0.5412.
The grader also says Code Cell UNQ_C8: Function 'c_statistic' is incorrect. Check implementation. and the point is not given.
The answer given here does not sound right, and there is a real issue in the way I’ve implemented my code, but I can’t see it…
Could you maybe help shed some light on this @Mubsi ?
1 Like
I identified my mistake thanks to @Mubsi. c_for_benefit is built to take as argument:
pairs (list of tuples): each element of the list is a tuple of individuals, the first from the control arm and the second from the treatment arm."
and I was doing the opposite when building pairs at the end of c_statistic… 
P.S.: my previous post has a typo and should read:
I get to an output of 0.4 instead of the expected 0.6.
2 Likes
Same issue here. I’m getting 0.4 not 0.6.
Hi @neurofire80,
How have you zipped the tuples, the order matters, if you have zipped in different order then it have shown wrong output. While filtering the tuples your use as x[2] == 0 for untreated patients and x[2] == 1 for treated patients is only true if you have zipped it correctly. Provide me your order of zipping pred_rr, y, w. There might be the problem present.
Regards,
Harsh Der.
I am getting the same issue.
tuples = [(i,j,k) for i,j,k in zip(pred_rr, y, w)]
# Collect untreated patient tuples, stored as a list
untreated = [x for x in tuples if x[2] == 0]
# Collect treated patient tuples, stored as a list
treated = [x for x in tuples if x[2] == 1]
This should be correct, however the test fails. When I flip untreated to X[2] == 1 and treated to x[2]==0, the test passes.
I also have the same issue of getting 0.4 instead of 0.6. So is the expected answer incorrect?
hi @ahwtsang
can you create a new topic. you have posted your issue on a very old thread.
if you check through other learner’s comment you will know why they got 0.4 instead of the expected 0.6. they have either mixed up how they built the pairing or other part of codes.