10
$\begingroup$

Can anyone tell me what is the name of this type of diagram ( if any )? Also can anyone suggest any tools, however simple, to plot such a diagram?

enter image description here

$\endgroup$
4
  • 1
    $\begingroup$ ok for anyone else that might see this... while the other suggestions are good in a general sense i have found a very handy piece of software exactly for this purpose... its just called treemap... it looks kinda antiquated but its really brilliant... here is the link treemap.com $\endgroup$ – lone_wolf32 Jul 23 '15 at 16:49
  • $\begingroup$ For a similar sort of display (but not quite the same thing), see mekko charts. $\endgroup$ – Glen_b Aug 3 '15 at 4:47
  • $\begingroup$ Although some people have argued that this is not a "mosaic plot," quite clearly it is, at least according to many authorities. The fact that some software might call it one thing and other software another does not make either one uniquely correct. $\endgroup$ – whuber Aug 3 '15 at 17:51
  • $\begingroup$ i also found you can do this in google sheets in google drive... i guess its pretty recent it wasnt there a while ago $\endgroup$ – lone_wolf32 Aug 11 '15 at 4:26
13
$\begingroup$

This plot called as "Tree plot" in tableau. You can see here to know how to do that. You can find trail version of Tableau here

Hope this helps !

$\endgroup$
0
5
$\begingroup$

If the difference between tree plot and mosaic plot coincides with the distinction between hierarchically-arranged categories and how single categories are broken down, then OP's image would appear to be a tree plot.

At first blush, I believed that the plot was a mosaic plot, which is one way to present stratified categories. A tutorial on the construction of mosaic plots in R can be found here.

I'll research this issue further when I have a moment.

$\endgroup$
7
  • $\begingroup$ According to the Wikipedia article, the purpose of a mosaic plot is to show relations between two or more qualitative variables. The plot asked about by the poster rather visualizes hierarchical discrete distributions. It looks similar, and there may be cases where both coincide, but I don't believe your answer is generally correct. $\endgroup$ – A. Donda Jul 23 '15 at 17:24
  • 2
    $\begingroup$ I think it is more accurately called a Treemap $\endgroup$ – llimllib Jul 23 '15 at 18:02
  • $\begingroup$ I've downvoted, as this is not correct. This is a type of diagram called a treemap, originally created by Ben Schneiderman. Here is one R package to calculate them, treemap, but I imagine there are others. Mosaic plots have a hierarchy based on the abscissa of the plot, no such hierarchy exists in tree map layouts. (Mosaic plots are more like stacked bar charts, it is just the bars have varying widths.) $\endgroup$ – Andy W Jul 23 '15 at 18:19
  • 2
    $\begingroup$ On the contrary, @AndyW, this answer is correct. It is well-supported by many references. Being called one thing by one reference does not make all other names wrong. $\endgroup$ – whuber Aug 3 '15 at 17:53
  • $\begingroup$ I disagree @whuber, and here is my response. I'm not quite sure where/why you think this is "well supported by many references" - care to elaborate? $\endgroup$ – Andy W Aug 4 '15 at 18:27
4
$\begingroup$

Treemapping is an information visualization method displaying hierarchical data by using nested rectangles.

mosaic plot Mekko chart or Marimekko chart

It has origins in mosaic plots (left) or Marimekko charts (right), adding nesting or embedding to the standard mosaic structure. The one you displayed is well-balanced, without elongated, skinny rectangles, that degrade the appearance of some treemaps generated by the "slice-and-dice" tiling algorithm.

So it belongs to the subspecies of "squarified treemaps", that can be decorated with colors or shading in cushion treemaps. A description is given in Mark Bruls et al. (2000) Squarified Treemaps, Proceedings of the Joint EUROGRAPHICS and IEEE TCVG Symposium on Visualization in Amsterdam, The Netherlands, May 29–30, 2000. Apparently, the idea of squarifying, i. e. constraining rectangles to low aspect ratio, was already present in M. Zizi M. and M. Beaudouin-Lafon (1994). Accessing Hyperdocuments Through Interactive Dynamic Maps, Proceedings of the 1994 ACM European conference on Hypermedia technology.

$\endgroup$
2
$\begingroup$

It looks like a variation of mosaic plot.

enter image description here

(source: statmenthods.net)

$\endgroup$
11
  • $\begingroup$ Tim, both your illustration and that in the question are clearly the same type of thing. Moreover, they both "show relations between two or more ... variables" by visualizing a certain hierarchy of conditional probabilities. Thus I think Andy W. and A. Donda were wrong to convince user777 that his answer is incorrect. I hope you will reconsider undeleting your answer. $\endgroup$ – whuber Aug 3 '15 at 17:49
  • $\begingroup$ @whuber since it is disputable, let it be so. I do not really consider this kind of plots useful and do not use those much, so I won't argue that I am right about this, but still it seemed plausible. $\endgroup$ – Tim Aug 3 '15 at 17:52
  • 2
    $\begingroup$ Disputation in this case is easily resolved by appealing to authorities. The briefest glance at the images Google displays in response to a search for "mosaic plot" is enough to show that a very large number of people consider the OP's example to be a mosaic plot. It's not hard to find some credible authorities among them. What really is of interest is not what various people might choose to call such a plot, but how it is constructed, how one reads it, and what it is good for. That is the kind of information I hope will eventually be included in some of the answers here. $\endgroup$ – whuber Aug 3 '15 at 17:55
  • $\begingroup$ I've posted a response to why the OP's chart is not a mosaic plot. How it is constructed is the key. In the sense that both treemaps and mosaic plots represent probabilities by areas they are the same, but one is not a variation of the other. Each have different layouts that make them visually distinct, and are used in different situations. $\endgroup$ – Andy W Aug 4 '15 at 18:31
  • 1
    $\begingroup$ I share the confusion in the opposite direction, I don't see how anyone would confuse the two. I've provided references in the linked separate page on the different layout algorithms. The squarify treemap layout in particular is easy to spot, it is intentionally designed to make the aspect ratio's of the boxes close to 1. A key point for mosaic plots is that it is still meaningful to make length comparisons in at least one of the dimensions. This is not the case for treemaps, only area of the boxes matters for treemaps. $\endgroup$ – Andy W Aug 4 '15 at 20:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.