# Computing questions: ggplot2

Pursuant to the recent apparent consensus (How much programming here?) that this site should accept some questions on programming statistical computations, I would like to inquire whether the site would be willing to take over the Q&A role currently served by the ggplot2 mailing list.

That is, ggplot2 is the latest (and greatest, in my opinion) 2D visualization package for R, and the mailing list has almost 1000 subscribers, leaving the original ggplot2 maintainer unable to give his full attention to all posts (though there are others that have acquired expertise in ggplot2 that help out regularly). It has been suggested (link to ggplot2 mailing list) that a new forum for obtaining help with ggplot2 might be reasonable, and that either stackoverflow or a proposal for ggplot2's own stackexchange might be a useful solution.

However, seeing that stats.stackexchange.com appears to be open to programatic questions, and anticipating that this might harken a move of most R-related questions from stackoverflow to stats.stackexchange.com, I wonder if a third option for providing help with ggplot2 might be to let ggplot2 questions be posted here.

Note that I'm not the head of the ggplot2 list; I simply thought I'd explore the acceptability of this option with you folks before reporting back to the ggplot2 list.

Note also that this may also relate to the questions (What to do with pure visualization questions?,Is it appropriate to ask for suggestions/solicit ideas on data visualizations?) on whether queries for advice on data visualization approaches are appropriate.

• This is a very old question (the 296th question on meta when CV was, I believe, still beta) that got bumped because of some tag-edits. Does anybody know the current status and whether anything happened as a result of this post? I see that the current mailing list is still active, but it is full of programming and helpdesk related questions like 'how to get two plots in one plot', 'how do I change the color', etc. (and this would nowadays be considered very off-topic). Should this question maybe get closed (if we ever do this on meta)? – Sextus Empiricus Apr 4 '19 at 11:34
• @MartijnWeterings, it's perfectly fine for Cross Validated to have evolved, &/or for people to have changed their minds & want to update this thread, but that doesn't make this off topic. – gung - Reinstate Monica Apr 5 '19 at 11:32
• @gung I wasn't sure about closing, but I pressed the button to close it because people might notice it better than just the comment. I have no problems with discussing ggplot. But the problem with this question is, I believe, that it is confusing. The information in the question, about the status of the mailing list, might not be correct anymore. In addition, people that hit on this question might not see at first sight that this is a very old question and they may start answering it with a much different perspective and not really answering the question that was the OP's original intent. – Sextus Empiricus Apr 5 '19 at 11:41
• @MartijnWeterings I wrote a short answer referring to the current status. If it gets sufficiently upvoted and/or accepted, then I think the problem with this thread will be solved? – amoeba Apr 5 '19 at 12:51
• Hi @Mike! Do you want to consider accepting my answer to mark this thread as "resolved" or do you want to leave it unaccepted because you disagree or think it should be extended etc.? Let me know. Cheers. – amoeba Apr 9 '19 at 10:51

As per https://stats.stackexchange.com/help/on-topic, the questions about how to achieve this or that in ggplot2 are (now, in 2019) off-topic and should be asked elsewhere:

Programming, ask on Stack Overflow. If the language is statistically oriented (such as R, SAS, Stata, SPSS, etc.), then decide based on the nature of your question: if it needs statistical expertise to understand or answer, ask it here; if it's about the implementation of an algorithm, routine data processing, or details of the language, then please refer to the collection of links to resources we maintain.

A good place to ask questions on ggplot2 is StackOverflow that has [ggplot2] tag with 27k+ questions: https://stackoverflow.com/questions/tagged/ggplot2.

This is a tricky question.

While I have not thought through the issues involved I am not sure if we should become a support site for statistical software. I am not sure I like to see all types of SAS, SPSS, MATLAB, R .... support questions coming to this site wholesale. For example, if posts on the R mailing list were to be posted here they are likely to drown all other questions. I could use tags to filter out the support questions, but still...

But, at the same time our current policy allows for computing questions. So, the above feeling of mine (not being a support site) is inconsistent with our current policy.

I am not sure how or where we should draw the line. Do we allow support questions for open source support software but not allow questions related to commercial software? I am not sure.

• Ah, I hadn't thought about the prospect of the R-help mailing list moving here. I don't think it's likely for the R-project folks to do it themselves, but I wouldn't put it beyond users to unofficially move. Indeed, there's already a lot of R questions on stackoverflow, so I anticipate that if there were a more stats-sounding site that was serving this same purpose, you'd get a lot of those users too, which would change the content here quite substantially. Hm, maybe "statistical computing and visualization" needs it's own stack proposal? – Mike Lawrence Aug 13 '10 at 18:23
• -1 because this answer is obsolete: our current (2019) guidelines regarding what is on-topic explicitly do not allow such questions. – amoeba Apr 5 '19 at 12:53

I have no problem with ggplot2 questions. I take Srikant's point about being overwhelmed by R or ggplot2 questions, but let's not forget that all the people doing R and ggplot2 programming are doing statistics/data analyses and so could add to the this site.

At present if someone asked a tricky ggplot2 question I would probably direct them to SO or the mailing list - but only because they would get a better answer.

• -1 because this does not conform to our current (2019) guidelines on what is on-topic. – amoeba Apr 5 '19 at 12:52

One of the nice things about this site as opposed to a mailing list is the natural language search that happens automatically as you formulate a title for your question. So long as we enforce appropriate tagging, so people can filter out questions they aren't interested in, and punish duplicates so we aren't overwhelmed, I see no problem taking ggplot2 questions here. The key issue to keep in mind is that we are not an appropriate place for bug reports, feature requests, or discussions thereof. Questions about ggplot2 (or other visualization software) asked here should stick to a simple: my data is this, I want it to look like this, how do I do that type of question. Quickly the basics will be covered and the flow of ggplot2 questions will become manageable.

• " I want it to look like this, how do I do that" this sounds to me more like a programming task and belongs to StackOverflow. – Sextus Empiricus Apr 1 '19 at 14:00
• -1 because this does not conform to our current (2019) guidelines on what is on-topic. – amoeba Apr 5 '19 at 12:52

In my opinion it depends very much on the nature of the question. For example, if the question is about how to use a certain color palette, then I rather have that answer posted on Stack Overflow. It is just about how to use some API.

After the feedback from Nick Cox on my former answer, I am reconsidered my answer. I believe topics about visualization should be better dealt in the data science site, whereas the focus of cross validated should remain statistics. As Nick points out, talking about how to organize the data will tend to involve how to do that with some specific tool.

• I am not clear that "tidy data" for example is a suitable topic for CV. That's all about how you hold data and is not intrinsically statistical. If you say that, then you're on a slippery slope to allow all sorts of questions about precisely how to reorganise data in R (or indeed any other software) of the kind that CV is reluctant to handle. A key criterion is: could this question be of use or interest regardless of what software people use? which rules out many of the questions that get posted here. – Nick Cox Apr 3 '19 at 8:20
• Sorry, I meant Cross Validated. And I agree with you on the "tidy data" topic. I see it would be more appropriate to forward that topic to the data science site. – jpmuc Apr 3 '19 at 17:46
• I disagree in part with your revision. A large fraction of statistics concerns visualization -- or graphics, as old-timers are likely to say -- and long may it flourish as a topic here. I just checked and found >2000 questions here on data visualization (and some may have easily escaped that tag) and >300 there on visualization (same comment). I need to look to see what I am missing there, but I don't think that any proposal that henceforth visualization questions belong there, not here, will work. Visualization is not just a popular topic; it is integral to most practice in statistics. – Nick Cox Apr 4 '19 at 7:09