9
$\begingroup$

Some datasets are perennial favorites. Would it make sense to have tags for questions about these? If I'm asking about a specific analysis I want to apply to a dataset, it might be helpful to see whether someone has done something similar to this specific dataset before.

I am thinking of tags identifying questions about

I thought about including a bullet point about the UCI ML data repository (currently 58 questions), but this would probably not be specific enough - better to tag specific datasets from the repository (if we decide to use such tags, and if the dataset in question is "notable enough").

| |
$\endgroup$
  • 2
    $\begingroup$ I can't see any point for that. I guess that the questions would either use the data as an example and the data itself would be irrelevant, or if asking about the data itself, the questions would be rather duplicates... (I may be wrong.) $\endgroup$ – Tim♦ Mar 23 '18 at 10:45
  • $\begingroup$ Questions duplicates of themselves? $\endgroup$ – Stephan Kolassa Mar 23 '18 at 10:47
  • 1
    $\begingroup$ Just to pick one of the first Iris questions, here is an answer that specifically addresses this dataset and its idiosyncrasies: stats.stackexchange.com/a/299646/1352 I'd say this argues for commonalities within the dataset that set questions about it apart. $\endgroup$ – Stephan Kolassa Mar 23 '18 at 10:50
  • 1
    $\begingroup$ We have tags for software packages which I know is not to everyone's taste so I would have thought dataset tags would be even more appropriate since datasets are the bread and butter of applied statisticians. $\endgroup$ – mdewey Mar 23 '18 at 13:20
  • 3
    $\begingroup$ I agree w/ @Tim. It's hard for me to see a question that is only about 1 of these datasets that would be on-topic. Typically, these are simply used to illustrate something else, where the something else is the point of the question. More generally, I think of tags as 1 of the primary ways we organize the information on the site (in keeping w/ our primary mission to create a repository of high-quality information about statistics via Qs & As). I can see someone wanting to learn about lme4, or using the tag to find a specific piece of information, but I don't see needing to learn about iris. $\endgroup$ – gung - Reinstate Monica♦ Mar 23 '18 at 13:44
  • 2
    $\begingroup$ It is rather telling that after almost eight years nobody who has asked a question has thought to tag it with the name of a dataset. (Although there is a kaggle tag now, it was created less than a day ago and only added retroactively to a few questions!) $\endgroup$ – whuber♦ Mar 23 '18 at 13:46
  • 1
    $\begingroup$ @whuber I can see reason for [kaggle] tag, since it hosts very specific kind of competitions, that need very specific kind of skills etc. $\endgroup$ – Tim♦ Mar 23 '18 at 14:03
  • 2
    $\begingroup$ @Tim My point is that nobody has thought it worthwhile to tag their own questions with dataset names. Moreover, there's hardly any need in the case of Kaggle: "Kaggle" will work fine as a search term; it doesn't need any semantic disambiguation. As such, one gets the impression that this thread concerns a non-issue. $\endgroup$ – whuber♦ Mar 23 '18 at 14:06
7
$\begingroup$

Well, judging from the comments, this is not a popular idea. (I'll self-accept this answer so the thread is closed.)

| |
$\endgroup$

You must log in to answer this question.

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