Quiz M7.04 Question 2 : Classifier's method should be call to get the predictions required by the function precision_recall_curve

Module 7. Evaluating model performance Classification metrics
#1

Excuse me, I have dificulties with the last question of the Quiz:
** Which classifier’s method should you call to get the predictions required by the function precision_recall_curve ?**
, because the documentation of de precision_recall_curve metric do not give me the information I need about the parameter probas_pred (possibilities). Could you please help me?

#2

Let me reformulate the question:

Given a classifier such as clf = LogisticRegresion(), once you have called clf.fit(X, y), which method(s) can you call such that its output can be provided to the parameter probas_pred of the function precision_recall_curve.

#3

I also found this question a bit confusing - “which method” instead of “method(s)” led me to interpret the question as looking for a singular response only.

3 Likes
#4

I agree with you @bweitz : the question lead to think we have to choose only one answer. In addition decision_function() was never described in the MOOC so far.
I propose strongly that the question has to be refomulated as the one that wrote @glemaitre58

2 Likes
#5

Had the same issue. Having an option with a function name that doesn’t exist, calling it with a syntax as if it were a (non-existent) function, is not cool. Bad question.

#6

in fact it was mentioned in the lecture “preprocessing for numerical features”:

This predictive pipeline exposes the same methods as the final predictor: fit and predict (and additionally predict_proba , decision_function , or score ).

but not explicited.

I take look in scikitlearn documentation but it s not clear what does that method. For LogisticRegression decision_score(x)

predict confidence scores for samples.
The confidence score for a sample is proportional to the signed distance of that sample to the hyperplane.

I confess i do not understand.

#7

Agreed, we will try to improve this question for the next MOOC session.

As a general comment, communication is hard I know but please try to stay constructive in your comments :wink::pray:! How to Contribute to Open Source | Open Source Guides is a great ressource about this kind of things.

@Lokijota for example rather than:

Bad question

Something like this feels a lot nicer:

This question is not great and there is some room for improvement here for example by doing this and this.

@echidne for example rather than

I propose strongly that the question has to be refomulated as the one that wrote @glemaitre58

Something like this feels a lot nicer:

@glemaitre58 wording would definitely have cleared up some confusion for me. This would be great to correct this!

#8

I do not understand what is not nice in my sentence. Here I used bold not to hurt your feeling but to insist on strongly. The poor refined phrasing I used is linked to my english level not in a will to hurt.
Belive me I’m a very nice person but it’s hard to speak like 2 old english women around a tass of tea. :wink:
BTW if you can explain that decision_function() that will be very nice of you.

#9

Belive me I’m a very nice person but it’s hard to speak like 2 old english women around a tass of tea.

This is a nice image I quite like it actually, maybe in my ideal world, the internet would have a lot more old ladys driking tea :grinning_face_with_smiling_eyes:.

Just to give you an idea maybe we have ~50 posts per day in this forum, mostly about problems that could be improved. This is very useful feed-back but it can be tiring after a while to see all these problems, especially given the time and energy we spent trying to make this MOOC as well as we could.

On your particular example when I see “I propose strongly” it sounds very demanding to me: it is easy for me to receive it as “stop immediately everything else you are doing to solve this problem because this is really bad”. I am exaggerating a bit of course but this is to give you some idea.

But yeah as I said communication is hard and the fact that all of us are not native English speakers does not help …

In general I feel like saying “here is the problem it would be great if it could be improved for example like this” feels a lot nicer than “this is bad please fix it!”. Not only it feels nicer but it is a lot more productive to keep the discussion moving forward.

2 Likes
#10

Hello. I agree with the fact that the question 2 is a bit misleading. I also thought that it had only one correct answer. I think it is especially the case for very debutant like me (also with basic Python level). In general in M7, there are few questions (1 or 2) and if we make a double mistake, there a risk to not validate several quizz. As it is hard sometimes to understand concept and some specificities when you are a debutant, it is difficult to know if a single answer is wrong because of it is alone or if it is the bad answer.
But I also know that writing quizz is difficult also !
Thanks for your job and I hope some of “student” remarks will help you.
Regards

#11

This quiz question has been removed.