How to ask a good question on data science SE?

Question

I have been browsing through the Data Science SE for last couple of months almost everyday along with other SE sites. I noticed that, in general, the questions asked on Data Science SE is often of poor quality. Many users are asking questions that are often too broad, just throwing out random idea and asking what others think and in most cases, they do not show any code or minimal reproducible example.

This is (mostly) not the case on stackOverflow questions (even in tags under R, Python questions which are the two main tools used by the community of DS SE).

The questions on data science are often difficult to ask because of issues like the nature of question (i.e. they are often not straightforward how do I do it using this language question) and reproducible codes may take some time to run due to the training process etc.

I am hoping that the community here can come up with a framework to ask great data science question. Hence, my question is:

How do I ask a good question on data science SE site?

For the sake of completeness, I should mention that I have read how to ask and asking better questions but none of them are tailored to the data science specific question.

I believe the answers should 'answer' following questions with good examples to follow:

1. How to be specific in asking a data science question?
2. How to produce a minimal working example for data science questions?
3. Is it possible to provide a complete list of dataset that may cover a broad variety of questions asking in future? If possible, new questions can pick one that suits them most and use that to produce MWE codes.

Answer 1

Although I don't know if the fraction of bad questions is unusually high, if it is, then it's probably because this is a topic with lots of newcomers learning on the job, and don't know what they don't know, and so have trouble articulating a problem in terms of art that veterans understand. They are more likely to not know how to break down a big question into something answerable.

What can be done?

• Have a look at the Cross Validated discussion: https://stats.meta.stackexchange.com/questions/1479/how-to-ask-a-good-question-on-crossvalidated
• Further clarify what the site is about, especially relative to SO and CV; we can actually edit the questions at https://datascience.stackexchange.com/tour for example
• Steer people to SSCCE as good practice for code-related and non-code-related questions, even: http://sscce.org/
• More moderators to provide more feedback (oh look, an election is underway)
• Give some examples here to stimulate a discussion about what good questions are like