Wikipedia talk:Graphs and charts

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Graphs are important for an encyclopaedia[edit]

I am surprised that this project page did not exist given the importance of graphs for an encyclopaedia. The probably should be a Wikipedia:WikiProject Graphs to coordinate the task of getting consistent looking graphs throughout Wikipedia. -- Alan Liefting (talk) - 07:32, 18 September 2008 (UTC)


I agree completely. I struggled for a long while to find any free application in MacOSX which is able to export charts into .svg. When i finally found something (https://www.gliffy.com/ , trial 15 days) now i do not know what colours or design types should i use to be compatible with wikipedia other articles.

Thinks i am missing:

  1. What colours should i be using, what is the accepted colour palette. (Consider accessible colors?)
  2. General guidelines what is the preferred style for organisations charts/tree charts/graphs of different type. (Rounded corners? Drop shadow? Gradient? How to name a legend/description?)
  3. There is talk about Vega-based graphing system, there is no mention anywhere what it is
  4. Maybe list of a) free, b) easy to use c) with SVG export capability applications for Windows/LinuxMacOSX. For ex. LibreDraw is not a possibility because of lack of export to svg, Inkscape on MacOSX depends on Xorg libraries which is not ideal from administrator point of view and also is not ideal for graphs. It was hard to find any easy to use free application with svg export - Regards , Patpi (talk) 08:07, 21 May 2015 (UTC)

Using on-the-fly generated graphics?[edit]

I just found the google chart project. Is it possible to use graphics generated by this engine without uploading it to commons? As an example I made a graphics on crude oil, one in german, one in english:

http://chart.apis.google.com/chart?cht=p3&chtt=Erdoel-%7CZusammensetzung&chs=400x150&chd=t:85,15,1,1,1&chl=Kohlenstoff%7CWasserstoff%7CSauerstoff%7CSchwefel%7CStickstoff&chco=606060,0060d0,d00000,ffff00,00c000 http://chart.apis.google.com/chart?cht=p3&chtt=Composition+of%7CCrude+Oil&chs=400x150&chd=t:85,15,1,1,1&chl=Carbon%7CHydrogen%7COxygen%7CSulfur%7CNitrogen&chco=606060,0060d0,d00000,ffff00,00c000

All information is present in the URL, from which the image is generated. There are much more types of graphics on the project page http://code.google.com/apis/chart/ which could ease keeping graphs, statistics, charts, maps, etc. up to date. -- Georg-Johann (talk) 14:21, 23 October 2008 (UTC)

Requested move[edit]

The following discussion is an archived discussion of the proposal. Please do not modify it. Subsequent comments should be made in a new section on the talk page. No further edits should be made to this section.

No consensus to move. Vegaswikian (talk) 17:58, 23 October 2011 (UTC)

Wikipedia:GraphsWikipedia:Charts – 19:04, 16 October 2011 (UTC)

  • Oppose up until today, this page was graph, the text said graph from its creation until this edit. A "chart" is a map. The edit should be reverted to the previous version and the associated page Wikipedia:Don't draw misleading charts should revert to Wikipedia:Don't draw misleading graphs that it was called until today. As no rationale was presented for any of this, I don't see why this should be left to stand. 70.24.251.158 (talk) 07:29, 17 October 2011 (UTC)
    • Wikipedia:Don't draw misleading charts has reverted to its old name Wikipedia:Don't draw misleading graphs . 70.24.251.158 (talk) 05:16, 19 October 2011 (UTC)
  • Comment I have informed the creator of this page, user:Alan Liefting about this request. 70.24.251.158 (talk) 07:32, 17 October 2011 (UTC)
  • Comment. Nominator needs to explain a sufficient reason for such a move request. Zzyzx11 (talk) 04:19, 18 October 2011 (UTC)
The above discussion is preserved as an archive of the proposal. Please do not modify it. Subsequent comments should be made in a new section on this talk page. No further edits should be made to this section.

Bars within bars[edit]

What if I want to put a bar or two in a diagram with several bars where I want to put information of what a bar contains?

As an example, in religion you have several sections of those religions.

Religion in Utopia 2019
[1]
Religious
  
90%
Christian
  
85%
Catholic
  
45%
Protestant
  
40%
Orthedox
  
15%
Other
  
10%
  1. ^ Cite error: The named reference 2019-census was invoked but never defined (see the help page).

Here I have put up 90% of the Utopia as religious in 2019, and of those religious, 85% are Christian, and of those, three religions are listed up as percentage of Christians and the remaining 10% are other religions or irreligious. I feel I should put that up in another fashion, but I'm not sure how. Jørgen88 (talk) 09:50, 5 October 2012 (UTC)

you could always add a horizontal separator and create a second graph with the break down of the denominations. Frietjes (talk) 17:02, 5 October 2012 (UTC)

Line graphs charts[edit]

How do I enter negative values in a line graph, or sub-integral units on the y axis? Say the range of values is −1 to +1, and I need lines every y=0.1. — kwami (talk) 22:07, 22 January 2013 (UTC)

are you asking about {{Line chart}} or <timeline>...</timeline>? Frietjes (talk) 22:46, 22 January 2013 (UTC)
Chart. Sorry. — kwami (talk) 23:04, 22 January 2013 (UTC)
there is/was a bug in that template when y_min is less than zero. I will see if I can fix it in a moment. Frietjes (talk) 00:32, 23 January 2013 (UTC)
I was able to fix the y_min less than zero bug, but the non-integer increment is going to be a problem
Integer increments Non-integer increments

EasyTimeline 1.90


Timeline generation failed: 1 error found
Line 22: ScaleMinor = increment:0.1 start:-1 gridcolor:scalesec

- Scale attribute 'increment' invalid. Specify positive integer.


Frietjes (talk) 00:41, 23 January 2013 (UTC)

Thanks! That solves the more important problem. — kwami (talk) 04:40, 23 January 2013 (UTC)

Creating a guideline for graphs and charts on articles[edit]

Closing per request at WP:ANRFC. The scope of this discussion focused on finding some common ground on factors that should be considered when using a graph to illustrate a data trend. So, to be clear, while this RfC is not going to define the exact guidelines surrounding graphs, it will illustrate some considerations:

  • MOS and guidelines for conventional images (e.g. WP:IMAGE, WP:OI) are not (currently) sufficient on their own to guide decisions in how we should make graphs.
  • Kerry_Raymond's idea of breaking this down was by 1) examining the source(s) from which the data came, and 2) examining the way the data are presented visually. This two-factor approach seems like a reasonable way conceptualize the issue. There was consensus that many relevant policies are already in place. As graphs take on a similar role as prose in that they present information/relationships/trends about a subject, policies regarding verifiability of the data, original research, and maintaining a neutral presentation of the data still remain relevant.
  • Citations for graphs are, by definition, necessary, as the data need to be verifiable.
  • Editors who create graphs that pull data from multiple sources need to tread carefully in that they should avoid manufacturing new arguments or comparisons not explicitly discussed in the source material. Editor meta-analysis (i.e. drawing your own conclusions from independent data sources) is textbook synthesis and is also not permitted. Graphs should not be constructed to imply such comparisons.
  • Some resources already written may be helpful in creating policy around visualization. Wikipedia:Don't draw misleading graphs and even our own article on misleading graphs are instructive. I, JethroBT drop me a line 17:56, 2 April 2014 (UTC)

This seems like a good start, but there are indeed other considerations (e.g. WP:RS was not really discussed here) that need to get fleshed out. A broader RfC will be necessary, and I'd recommend that a draft be created in preparation, so that the focus of the RfC can be on constructing the policy rather than on just the ideas. I, JethroBT drop me a line 17:56, 2 April 2014 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

We currently do not have a policy nor a guideline for charts and graphs. We are currently having a difficult time reaching consensus at Talk:Alejandro García Padilla due to several graphs being included on the article that portray the economic activity and employment during the governor's tenure. The same thing happened with Barack Obama and File:US EmpStatsBLS Jan09-Feb13.png where this graph was discussed at Talk:Barack Obama/Archive 77#Labor force participation rate. We do the same for George W. Bush and File:Deficits vs. Debt Increases - 2009.png. Unfortunately, as we don't have a guideline specifically for this, we are reaching a stalemate where neither side is right nor wrong.

We need to fix this problem.

We need to move things forward and establish a guideline on how graphs should be prepared; specifically regarding data trends and their sources. We should also take in consideration graphs prepared from data where the data is shown "as is" but the graph per se has not been published by a reliable source. In other words: we have the raw data, but then an editor creates a graph from such raw data even though no one else has published said graph before. For example, File:US EmpStatsBLS Jan09-Feb13.png was drawn from raw data extracted from [1] & [2] — nobody else in the world has published such graph except us, but the graph can be extrapolated from the raw data published by the reliable source.

This is the only page on all of Wikipedia where this subject is discussed, but as of today, it is merely a project page written back in 2008. Right now, our policies and guidelines do not establish how graphs should or can be prepared as an official guideline (this page is NOT official as of this writing). Finally, our policies and guidelines do not prohibit creating graphs from WP:PRIMARYSOURCES but several people are concerned on whether this should be done on WP:BLPs or not.

Relevant policies and guidelines for this discussion include:

Let's start a discussion on this RFC to reach consensus and set up the way for future editors.

The question is: how should graphs on Wikipedia be created?

Ahnoneemoos (talk) 17:49, 22 December 2013 (UTC)

  • this seems to be doing things the wrong way round. Policy is descriptive on Wikipedia. First we need to get consensus on the respective articles, and at some point we can then conclude that apparently one thing or the other is policy. We shouldn't be trying to create policy to fix the outcome of some discussion. 62.140.132.90 (talk) 08:42, 9 December 2013 (UTC)
    We really do not have a specific process to create policies and guidelines save that they have to be discussed, require community support, and be accepted as policy or guideline through consensus. From WP:PGLIFE itself, for your convenience:

[Other policies and guidelines] developed as solutions to common problems and disruptive editing.

Which is what we are trying to do here: having a discussion about this issue and find out what's the best course of action (whatever that may be; which as you have pointed out may be to do nothing about it).
We are not trying to fix the outcome of a discussion. We are trying to have a discussion about our current incongruences and lack of consensus across Wikipedia. Consensus on these matters seem to be local right now and we need a systemwide guideline that editors can point to and say: "oh, this is what our guidelines advise, therefore we should attempt to do things this way." (whatever that 'way' may be)
Ahnoneemoos (talk) 14:58, 10 December 2013 (UTC)
  • Comment: I am involved in several graph-related disputes currently percolating at Talk:United States which commenters here might want to review. I am struggling with how to interpret the WP:OI policy which states in part, "Original images created by a Wikipedian are not considered original research, so long as they do not illustrate or introduce unpublished ideas or arguments, the core reason behind the NOR policy. Image captions are subject to this policy no less than statements in the body of the article." Furthermore, there are some technical solutions which may be helpful at Wikipedia:How to create charts for Wikipedia articles (and it may be useful to link to the results of this discussion from there) as well as e.g. [3] and [4]. EllenCT (talk) 07:55, 19 December 2013 (UTC)
  • Comment: My graphic skills are way out of date, so I cannot comment on technical issues. However, I am interested and willing to comment on documentation and copyright related to graphs. We are in sore need of guidelines and policies related to documentation of graphs that are uploaded either here or Wikimedia. I am often reluctant to use and will sometimes remove data graphs from articles if the year and source of the graph or its primary data are not adequately cited. Meclee (talk) 22:01, 19 December 2013 (UTC)
  • Comment: I think it is important to use graphs (and other visualisations) because they are a powerful tool to enable people to understand a set of numbers, and we should not (from a policy perspective) be trying to do anything to discourage the use of graphs. There's two parts to what makes an acceptable graph, which I think can be decoupled. The first is the source dataset and the second is the visualisation of that dataset (as a graph or whatever format). It seems to me that the general principles for sources for article text appear equally to the source dataset: reliable, secondary and properly cited. The second question is whether the visualisation is "honest" (which we want) or "dishonest" (encourages misinterpreation). Generally visualisations are dishonest because people are trying to push a Point of View (e.g. "the rising cost of X during President Y's term"). A common way to be dishonest is to range your Y-axis over a narrow range of values which tends to make the lines on the graph look steep when they'd look flat if your Y-axis was over a a wider range of values. This is an area where some guidance is needed on how to ensure a visualisation maintains NPOV. There is probably a policy needed as to when/whether the original dataset should be cited in any WP article that uses the graph versus only be cited in the Commons entry where the graph (as an image) is held. We probably need some guidelines on manipulating data, e.g. adjusting for inflation. Obviously it often makes sense to do such data manipulations but I would suggest it should not be done as part of the creation of the visualisation itself but via an intermediate dataset (which should be available and appropriately described and citing the original dataset). That is, a visualisation should be a reasonable straightforward presentation of a set of data. Kerry (talk) 22:54, 19 December 2013 (UTC)
 :I agree with Kerry about the need to use graphs and the breakdown between source and visualization. It occurs to me that there are many guidelines and policies already in place that relate to these issues, but it is probably not clear that WP:CITE and WP:NPOV, for eample, should also be applied to graphics. That part of the problem might be solved rather quickly by creating a guidelines page for graphics and linking to WP guidelines and policies that also apply to graphics. Any novel agreed-upon additions could then be made as consensus is achieved. Meclee (talk) 23:42, 19 December 2013 (UTC)
  • Comment: requiring sourcing of data in graphs-- even graphs that do not include original research-- would require a caveat to be included in WP:OI. And I support that. Holdek (talk) 01:01, 22 December 2013 (UTC)
    Agree. I think the principles elaborated there are equally applicable to graphs, whether imported as images created in tools outside of WP (e.g. Excel charts) or created within Wikipedia syntax e.g. Line charts. "Show it like it is" would appear to be the starting point for visualisation and that manipulations are only acceptable if they are properly disclosed and are not Original Research or pushing a Point of View. Kerry (talk) 01:21, 22 December 2013 (UTC)
  • I don't know about this. This page is really about how to make a graph, e.g., "type this, then click on that...". I don't think that it would be helpful to cram in a bunch of stuff about when to use them and how to evaluate NPOV issues. Perhaps a completely separate page would be better. (And the IP might want to see WP:PROPOSAL for information on how to create guidelines. An RFC like this is a very reasonable step in that process.) WhatamIdoing (talk) 04:48, 31 December 2013 (UTC)
  • Comment One difficulty or problem I have seen is the combining of two or more statistics/studies into one graph. The underlying data is from peer-reviewed papers but combining the papers seems like synthesis or OR. It's an issue. Capitalismojo (talk) 05:31, 1 January 2014 (UTC)
    This should be no different than combining several sources in a paragraph with text; the mere juxtaposition of information is not original research, except when it implies a comparison that introduces a value judgement. If the graph is merely an accurate summary of the information in the references which is presented in a neutral way, there is no problem combining them. If the new graph advances a position not found in either source, it's synthesis and therefore original research. In brief: summarizing=OK, hinting at new implications=BAD. Diego (talk) 01:09, 4 January 2014 (UTC)
    P.S. There are two simple tests specific to graphics that can be applied to assess whether the image is problematic: 1) would the same information be synthesis if presented in tabular form? and 2) does the image have characteristics of a misleading graph? If the answer to both tests is no, the image is probably safe, provided it's subject to the other core content policies (neutrality, due weight, etc). Diego (talk) 01:23, 4 January 2014 (UTC)
  • I agree with the sentiment expressed above: that charts and graphs are more similar to prose than images, and that our image guidelines are not currently sufficient. Even picking the scale of the chart a certain way can mislead the reader. There are also issues with cherry picking a certain data series to make a point that borders on synthesis. For a great example, check out the history on this file File:Median_US_household_income.png. The original image was clearly chosen to make a point. Gigs (talk) 18:37, 17 February 2014 (UTC)
  • As noted by contributors above, charts and graphs are a bit more complex than just image guidelines. On one hand, as they are in format, images, they need to conform to WP:IMAGE, but as they convey information they obviously need to follow Wikipedia's conveyance of information policies and guidelines, such as WP:V, WP:NOR, and WP:NPOV. More specifically and in particular: WP:OI, and Wikipedia:Don't draw misleading graphs. I also think the guidelines noted at MOS:TABLES would be helpful as well. - jc37 23:27, 22 March 2014 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


Consider accessible colors[edit]

Please consider a section that talks about using accessible colors. People with color vision deficiency prefer clear colors with clear contrast. It might be prettier to use shading and gradients, but consider that some colors that look far apart for non-color vision deficient people look very close for others and gradients make it even more ambigious. Especially red and green are problematic. There it is best to choose high contrast colors but also use brightness and use full color intensity (instead of pale or dark colors). This is a big problem in Wikipedia in many articles and raising awareness of this would help a lot. Also remember that graphic representation makes things easier to understand, than text or symbolic representation. So colors should still be used, and not just some alternate symbolic coding (like numbers to point at sections of images). Ideally there would be a color contrast check, and a guide to choose appropriate colors that is shown before people use graphics or upload them. Thanks a lot. — Preceding unsigned comment added by 88.219.148.137 (talk) 09:21, 23 May 2014 (UTC)

Tech talk on the new graph system[edit]

Hello everyone,

There's an upcoming talk about the new Vega-based graphing system this week. It's 13:00 PDT on Thursday, 14 May, which I think works out to 21:00 UTC. If you're interested, the link is https://www.youtube.com/watch?v=j7DTn9jHnI0 That link might work afterwards, too; they often record these kinds of talks. There should be a demo as well as some technical information about how the system is built. Whatamidoing (WMF) (talk) 02:02, 13 May 2015 (UTC)