10
$\begingroup$

We have spent some time discussing questions that are not perfect. When a new user shows up, and doesn't know how to ask a question or what questions to ask, seeing lots of people talking about what doesn't belong can send the message, "Don't ask questions; you don't belong here."

So, let's:

  • identify some questions that have been asked so far during the beta that really seem in some way to be perfect for this site, and
  • discuss what it is that makes them good questions.

A set of great questions, with explanations for what makes them great, would be a very useful addendum to What topics can I ask about here? in the Help Center.

$\endgroup$

2 Answers 2

Reset to default
5
$\begingroup$

I'll start with the question that prompted this discussion:

https://datascience.stackexchange.com/q/528/322

This question comes from a new user without much participation in the beta so far. It's not a perfect question in every way, but I think it has two qualities that make it excellent for DataScience.SE:

  1. Immediately identifies a specific scope that is broadly relevant. Polling data is an area of much analytical interest, but the OP right away tells us to consider the case where there are more than two candidates and a small number of data points. Questions that identify this type of scope have the wonderful quality of of being interesting and relevant while still being answerable in the way that SE questions should be.

  2. Suggests a complete approach to tackling the problem. Some questions come in the form, "I have done A, B and C. What do I do next?" That often leads to grouchy comments and users talking about their favorite software. In this question, the OP tells us their goal and describes their plan, in terms of a recognizable algorithm and an established model. It gives other users something focused and concrete to respond to, independent of the details of implementation.

Any question with those qualities, even if it has issues with other things like language or formatting, has the core qualities of a great question on this site in my opinion. The little details can be fixed via community editing.

$\endgroup$
2
  • 1
    $\begingroup$ I see the question used as an example was migrated to CrossValidated after all. $\endgroup$
    – tomaskazemekas
    Commented Jun 29, 2014 at 8:09
  • $\begingroup$ @tomaskazemekas I think that's particularly unfortunate considering that no one publicly expressed a counter opinion, including whichever pro tem mod performed the migration. But I've been very disappointed in general by the lack of discussion on this topic. If no one else cares to identify one "great" question out of a field of over 150, that doesn't have great implications for the site. $\endgroup$
    – Air
    Commented Jun 29, 2014 at 17:40
3
$\begingroup$

I'm new the DS.SE, but have been using SO.SE for years. Having just now looked in the on topic questions, I find there aren't any. This may be part of the reason there is so much trouble in distinguishing between CV.SE and DS.SE.

I do see the need for DS.SE, but there is a danger it will never get past the beta.

I'm currently too low to edit the 'on topic' page.

Suggested questions

  1. When using Stream Insight module from Azure with 10 ms data from one process, what statistical tool should I use to best monitor a change of process - either distribution/mean?
  2. I would like to produce a infographic on the 'Brexit' referendum. What public data sources might I use and how to summarise in a meaingful way, in order to visualise into a dashboard?
  3. Is R or Python better for executing an ARIMA model over SPARK?
  4. How does the auto scale feature of Azure work in relation to Stream Insight / blob storage and does this improve the performance of Azure ML models?

Also, there are some very good questions on the DS Area 51 site, which could be considered for addition to the DS.SE on topic questions list. (there are also some very poor questions as well, which should have never been voted up, IMHO!)

$\endgroup$
4
  • $\begingroup$ Most pages in the Help center cannot be customized except by staff. The on-topic page can be customized by site moderators. $\endgroup$
    – Air
    Commented Mar 22, 2016 at 16:05
  • $\begingroup$ That's perfectly fair. I like the way that people who invest in the community grow into higher permissions of access ... Perhaps someone with permission could edit the on topic page to add questions that have some value to the community? $\endgroup$
    – Marcus D
    Commented Mar 22, 2016 at 16:12
  • $\begingroup$ That sounds like a good suggestion to me. The most expedient way to promote such a proposal would be to give it a brief write-up as its own question for discussion here on Meta. New questions on Meta prompt an inbox notification for moderators. $\endgroup$
    – Air
    Commented Mar 22, 2016 at 16:16
  • 1
    $\begingroup$ Sounds good ... I'll repeat this answer as a new question ... thx $\endgroup$
    – Marcus D
    Commented Mar 22, 2016 at 16:21

You must log in to answer this question.

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