Data Science is a pretty broad subject and defining what is on topic or not seems quite hard. This difficult is reflected on our help-center that does not specify what is on topic but rather give few examples of on-topic questions.

I think we should make a community wide effort to define the "on-topic" on help-center. And we can go by this in different ways:

Democracy and Data Science

  • Could we implement a new temporary feature to vote the questions as on and off topic to avoid vote-downs for off-topic since it damages reputation and that is unfair given that this definition is a work in progress?

  • Could/Should we use this as an opportunity to create a NLP algorithm/dataset to decide if a question is or isn't on topic for DS SE?

Democracy by vote

  • Can we create featured topic where we can propose definitions for on-topic and off-topic questions on meta and use the voting process to decide "on-topicity"? If so, answers to that topic should cover one and only one definition and should be made community wikis so they can be easily improved.

2 Answers 2


I think the on/off-topic examples could be improved (though I'm not sure how to change them). No matter what, on/off-topic is subjective at the edges, and won't be perfect. That's OK. In practice, I find most questions here are on-topic according to my understanding so it must not be that hard to intuit when this is the right place to ask.

Closing a question doesn't affect reputation, so I don't see an issue there. Mods don't just close questions, they migrate them to the 'right' place.

People shouldn't be downvoting just because something is off-topic. I'm sure not every person does that 'correctly' but is it a significant problem? That is, it's already supposed to work in the way you suggest.

I don't think an ML model can sort this out better, and don't think anyone is going to build that. (Can't speak for StackExchange though)

Yes you can open questions about what's on-topic here and we can vote on them. I just don't know how one changes the examples in the SE documentation.


Example of answer that would fit the second proposal

On-topic: Questions about Image Processing explicitly for Data Science applications:

  • Feature engineering and feature understanding
  • Pre-processing steps
  • Similarity measures for images

Examples of questions that would fit to this definition:

Examples of question that do not fit this definition:


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .