11
$\begingroup$

Some questions, often from new users, pose a practical problem of modeling or forecasting a behaviour of some system and ask for suggestions for modeling or forecasting techniques. In some cases the choice is really very broad, and I start thinking whether the question itself is not too broad (a reason for closing).

However, the question is often specific; there is a concrete problem. But still there are so many possible solutions to be offered.

Should such questions be left intact? Closed? Edits be suggested?

$\endgroup$
  • 8
    $\begingroup$ I would start by commenting to see if you can engage the OP & get the Q to the point where it would be answerable. Although I don't follow forecasting, eg, I sometime see them, & my impression is you are trying that already & doing a good job of it. $\endgroup$ – gung Jul 27 '16 at 21:00
  • 1
    $\begingroup$ What I do with this kind of questions is either provide brief comment or answer pointing some general direction to go (where to start, references), or vote to close as too broad. In many cases I do both. $\endgroup$ – Tim Jul 29 '16 at 11:08
  • $\begingroup$ @gung, thanks. This takes a lot of work and therefore makes the question less attractive from an answerer's point of view. $\endgroup$ – Richard Hardy Jul 29 '16 at 11:29
  • 1
    $\begingroup$ @Tim, that sounds fine, it is just that I am not sure whether the reason for closing is sound. $\endgroup$ – Richard Hardy Jul 29 '16 at 11:29
  • 2
    $\begingroup$ It is a lot of work--I do a fair amount of it myself--but you do a good job of it & it's important. $\endgroup$ – gung Jul 29 '16 at 12:10
  • $\begingroup$ @gung, fair enough. I appreciate the encouragement. $\endgroup$ – Richard Hardy Jul 29 '16 at 12:15
5
$\begingroup$

One option is to provide a list of potential approaches and a corresponding set of general resources in meta - much as we do for questions about software (one package with links per answer); in this case one methodology with resource links per answer.

Then when we go to close as too broad we can point directly to that assistance.

It's a fair bit of work to do in one go but by doing one-per-answer it can be done piecewise - list a few major things, do those and then work down to other methods over time, as we see there's value in doing so. Such answers can be done by people familiar with the approaches they're writing about.

$\endgroup$
  • 3
    $\begingroup$ I wouldn't be surprised if there is a canonical answer somewhere on the main site that at least partially accomplishes this (maybe one of the suggested reading material). $\endgroup$ – Andy W Aug 5 '16 at 13:04
  • 2
    $\begingroup$ Why not put such a thread on the main site? $\endgroup$ – gung Aug 5 '16 at 13:25
  • 3
    $\begingroup$ I agree with @gung. In fact, Richard can ask and self-answer a question calling it "A Primer on Forecasting Techniques" (or a survey). If done well, this thread can become popular and we can use it to close other questions as duplicates. That's the benefit of having it on the main cite. [Disclaimer: I am not following our forecasting questions at all, so have no idea how feasible/useful this will be in practice.] $\endgroup$ – amoeba Aug 5 '16 at 14:41

You must log in to answer this question.

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