6
$\begingroup$

@gung recently created the tag which now has 9 questions: 2 old threads tagged by @gung and 7 new threads (plus 1 migrated to SO). I have two questions about the tag:

  1. Do we need it?
  2. If we need it, can we at least replace pcf with pacf in the tag name?

Edit: 2. Has been done already, only 1. remains relevant.

Regarding 1., I think we were doing more or less OK without it. Most of the questions containing ACF and/or PACF I have encountered on CV would ask for

If we were to get rid of the tag, it could be

  • made a synonym of or
  • just replaced by the above-mentioned tags in the few threads tagged with but not the ones above, and deleted in other.

Regarding 2., PACF is the standard acronym of partial autocorrelation function in time series econometrics, while I cannot remember seeing PCF before (though I would not be surprised it exists). On Cross Validated, PACF can be encoutered 670 times while PCF only 17 times. I think having PCF in place of PACF in a tag name may be confusing and may spread the unwarranted use of the nonstandard acronym (so that instead of searching for PACF one will now have to search for both PACF and PCF to capture all the relevant posts).

$\endgroup$
  • 1
    $\begingroup$ I don't have an opinion here, but FWIW when gung created this tag, he seems to have put it on two threads: stats.stackexchange.com/questions/106038 and stats.stackexchange.com/questions/108983. The other 8 threads with this tag appeared later. Most of them do not have [autocorrelation] tag. Would your suggestion then be to make [acf-pcaf] a synonym of [autocorrelation]? $\endgroup$ – amoeba says Reinstate Monica Jun 19 '18 at 12:49
  • 1
    $\begingroup$ @amoeba, this could be logical. Either we keep it (then I would urge to change the name to acf-pacf) or make it a synonym, and then the intuitive synonym for me is indeed autocorrelation. $\endgroup$ – Richard Hardy Jun 19 '18 at 13:04
  • 1
    $\begingroup$ Seeing that many people started using this tag after it was created, I think having it around is a good idea. The question remains whether to make it a synonym or not (on which I have no opinion). $\endgroup$ – amoeba says Reinstate Monica Jun 19 '18 at 13:57
  • $\begingroup$ 7 questions over 2 months (besides the two old threads @gung tagged himself when creating the tag) – not too intense, but not dead either. $\endgroup$ – Richard Hardy Jun 19 '18 at 14:02
  • 1
    $\begingroup$ OK, the tag's name has been changed to acf-pacf. $\endgroup$ – gung - Reinstate Monica Jun 19 '18 at 17:33
  • 1
    $\begingroup$ I seem to have a vague memory that part of the point was to make it a synonym, but I'm not really sure. Couldn't someone have a question about the plots w/o it being a question about autocorrelation? What if someone wanted to ask, 'how to identify the MA parameter from this ACF plot?', should the system retag that (silently) to [autocorrelation]? $\endgroup$ – gung - Reinstate Monica Jun 19 '18 at 17:43
  • $\begingroup$ @gung, good example. Perhaps it shouldn't. $\endgroup$ – Richard Hardy Jun 19 '18 at 17:53
  • 2
    $\begingroup$ ACF and PACF are typically used in identifying ARIMA orders in a Box-Jenkins type approach. Looking (only) at the titles of the acf-pacf questions, it seems like most of these could be retagged box-jenkins. To be honest, I have little preference either way, since Box-Jenkins is really not the modern way to go about ARIMA model selection - nowadays, one uses information criteria. $\endgroup$ – S. Kolassa - Reinstate Monica Jun 21 '18 at 0:43
5
$\begingroup$

(Turning my comments into an official answer.)

I changed the name of the tag to .

I am open to making it a synonym, but I'm also not convinced it should be a synonym of [autocorrelation] per se. I don't know enough about the topic to be confident of making any definitive statements, but it seems to me that someone might want to ask a question specifically about the plots themselves and that allowing the organization of such material by having the tag can be useful. In particular, they aren't really synonyms, in the literal sense. That doesn't mean that they can't be made into tag synonyms, but does suggest we should think it through first. For example, I could imagine someone asking, 'how to identify the MA parameter from this ACF plot?'. I wouldn't want such a question to be silently retagged to .

Moreover, as @amoeba notes, the tag has been used by others even in the short time since I created it. That implies there may be some demand for it.

Thus, while I remain open to making it a synonym of something, I lean towards leaving it alone for now.

$\endgroup$
  • $\begingroup$ +1 from the OP. $\endgroup$ – Richard Hardy Jun 20 '18 at 14:16
  • $\begingroup$ @RichardHardy Is the issue settled then? If so, perhaps accept gung's answer? $\endgroup$ – amoeba says Reinstate Monica Jun 27 '18 at 14:22
  • $\begingroup$ @amoeba, good idea. $\endgroup$ – Richard Hardy Jun 27 '18 at 14:23

You must log in to answer this question.

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