ggplot2 3D Bar Plot

Question

I Know this sounds basic, but have a been searching for literally more than an hour now without success. I'm simply trying to plot a 3D bar plot in 'R' using the 'ggplot2' package. My dataframe looks something like this:

 x   y     z
t1   5   high
t1   2   low
t1   4   med
t2   8   high
t2   1   low
t2   3   med
t3  50   high
t3  12   med
t3  35   low

and I want to plot something like this on it: enter image description here

Any help is more than appreciated!!

You should realize that R aficionados think of Excel as a rich source of examples of how _NOT_ to do things (properly) and that 3D barplots are one of the most deprecated strategies for communication of accurate quantitative information. — IRTFM, Nov 07 '14 at 04:52
@BondedDust ha! yes, i do realise that. probably the reason why i cannot find any examples at all. but i still want to try, and any pointers from you will help… i just need to know how to add a third axis to my `geom_bar()` — Tavi, Nov 07 '14 at 04:55
This may be an avenue worth investigating: http://stackoverflow.com/questions/23261760/how-to-generate-3-d-bar-graph-in-r — Docconcoct, Nov 07 '14 at 04:56
@Docconcoct thanks, but all the responses are for 2d plots!! — Tavi, Nov 07 '14 at 04:59
From the author of rgl: https://stat.ethz.ch/pipermail/r-help/2007-September/141578.html and from @hadley: http://vita.had.co.nz/papers/prodplots.pdf — IRTFM, Nov 07 '14 at 05:02
I don't think you can make a 3D barplot with `ggplot2`. Like exceeding the speed of light in our universe, there are just some things that are forbidden by the fundamental laws of the Hadleyverse. In any case, a line plot grouped by color will be much more informative. For example, with the data you listed above try this (where I've named the data frame dat): `ggplot(dat, aes(x, y, group=z, colour=z)) + geom_line() + geom_point()`. — eipi10, Nov 07 '14 at 06:00
@eipi10 Could not agree more. This comment of yours is actually worth posting as a decent answer IMO. — tonytonov, Nov 07 '14 at 07:15
Does it have to be `ggplot`? (If not, take a peek at the examples in `?latticeExtra::panel.3dbars`. — jbaums, Nov 07 '14 at 07:36
@jbaums thank u J I'll check it now. No it doesn't have to be ggplot. Anything I can do in R would be just fine :) — Tavi, Nov 08 '14 at 14:35
@eipi10 thank you I'll try that in a second and post it as answer possibly;) — Tavi, Nov 08 '14 at 14:36
@tonytonov good to know I haven't tried it just yet as I've been away but will do in a second!! — Tavi, Nov 08 '14 at 14:37

score 24 · Accepted Answer · answered Nov 08 '14 at 21:34

24

As mentioned in comments, 3D plots usually aren't a good choice (when other options are available) since they tend to give a distorted/obscured view of data.

That said, here's how you can plot your data as desired with latticeExtra:

d <- read.table(text=' x   y     z
t1   5   high
t1   2   low
t1   4   med
t2   8   high
t2   1   low
t2   3   med
t3  50   high
t3  12   med
t3  35   low', header=TRUE)

library(latticeExtra)

cloud(y~x+z, d, panel.3d.cloud=panel.3dbars, col.facet='grey', 
      xbase=0.4, ybase=0.4, scales=list(arrows=FALSE, col=1), 
      par.settings = list(axis.line = list(col = "transparent")))

enter image description here

answered Nov 08 '14 at 21:34

jbaums

27,115
5
79
119

1

@maryam - just noticed the levels of `z` are ordered incorrectly. You can correct that by explicitly stating the level order when creating the factor. – jbaums Nov 09 '14 at 09:17
oh yeah, it goes med -> low -> high instead of high -> med -> low thanks for the correction :) – Tavi Nov 09 '14 at 21:27
What does `y~x+z' mean? – DavideChicco.it Apr 09 '15 at 20:42
@DavideChicco.it - it says to plot `y` against `x` and `z` (where these are elements of data `d`). – jbaums Apr 09 '15 at 22:00
@jbaums Thanks, I should have read the documentation. Another little help: how could I increase the dimensions of the sides of the parallelepipedon? Thanks! – DavideChicco.it Apr 10 '15 at 14:31
5

The code raises error : Error Using Packet 1 Non-numeric argument to binary operator – Matthew Son Jan 20 '22 at 21:07

score 0 · Answer 2 · answered Jan 08 '23 at 15:33

The {rayshader} package is another package that allows 3D visualisation - according to the author, it will only allow 3d visualisation for data that has indeed three dimension. Here based on a tile plot.

NB: the below required for me to install current GitHub versions of both the rayshader (v.0.33.3) and rgl (v.1.0.0) packages.

# devtools::install_github("tylermorganwall/rayshader") ## NB this will install a lot of dependencies
library(rayshader)
library(ggplot2)
d <- read.table(text=' x   y     z
t1   5   high
t1   2   low
t1   4   med
t2   8   high
t2   1   low
t2   3   med
t3  50   high
t3  12   med
t3  35   low', header=TRUE)

p <- ggplot(d, aes(x, z, fill = y)) +
  geom_tile() +
  scale_fill_fermenter(type = "div", palette = "RdYlBu")

plot_gg(p)
render_movie(filename = "plot.gif")

ggplot2 3D Bar Plot

2 Answers2

Linked

Related