Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Schemes and graph combination rethink?

From   Adrian Mander <>
Subject   RE: st: Schemes and graph combination rethink?
Date   Fri, 27 Oct 2006 12:57:21 +0100

I agree that my wants might be in the minority but it didn't take much to break
graph combine and I personally think that this needs a rethink.

I could write the command that I want by constructing the marginal boxplots
from scratch myself and placing them where I want them but I think graph combine
could be more useful.

I think one possible solution is shown below
that keeps Stata's liking for relative scaling but fixing the plot region box.
Then we would have control over of exactly where a plot region is and let the graph region scale
around this box. This shouldn't be too difficult to implement not sure if it breaks lots of other
programs though

(10%,20%) (90%,20%)
| |
| |
| |
| |
(10%,80%) (90%,80%)


At 12:21 27/10/2006, you wrote:

My understanding is similar to Svend's.

Otherwise put, you want -plot combine- as well as
-graph combine-. I suspect that would be a lot of
work to write and is really needed by only a few
people, and so it is not likely to be high on the agenda.

Another kind of solution is to fix the problem upstream.

I've found that a -twoway- with

plotregion(style(none)) yscale(off) xscale(off)

clears the decks, after which you can put what
you like where you like with a series of individual
calls. That's how I wrote programs to do triangular
and circular plots, neither of which is, on the face
of it, a -twoway- plot at all.

I quite like marginal box plots, but I like marginal
rugs (ticks at every data value) even more. Putting
minor ticks at every data value after calls to -levelsof-
is one way to do that.


Svend Juul

> Adrian wrote:
>     gr_example2 combine5
> is an example in the -graph combine- help file
> note that -gr_example2- is the command and combine5 is the option
> The rotation of the labels actually changes the size of the histogram
> plot region and hence the graphs do not slot together exactly (as it
> did before)
> --------------------------------------------------------
> Now I see - partially, at least (the well hidden command names inside
> the help file still puzzle me).
> Your main point, I think, is that you can only control the plot area
> size and location indirectly. It is the graph area minus what happens
> to be outside the plot area, and that can give troubles or
> inconveniencies; combined graphs is one of the cases.
> I agree, but I believe that there is not much hope for a change.

*   For searches and help try:
*   For searches and help try:

© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index