Message-ID: <18379.64319.567812.660719@stat.math.ethz.ch>
Date: 2008-03-03T13:21:03Z
From: Martin Maechler
Subject: Make plots with GNUplot. Have anyone tried that?
In-Reply-To: <30c82f5f0802291554p1b03cbd4p4b788a296191ec21@mail.gmail.com>
>>>>> "LH" == Louise Hoffman <louise.hoffman at gmail.com>
>>>>> on Sat, 1 Mar 2008 00:54:56 +0100 writes:
>> If you still want to then read ?write.table, that can export your data
>> into a spreadsheet-like ascii format which can be used from GNUplot
>> easily.
LH> Very interesting.
LH> So if I e.g. write:
LH> ts.sim <- arima.sim(list(order = c(1,1,0), ar = 0.7), n = 200)
LH> ts.plot(ts.sim)
LH> How do I know the names of the rows to put in the data.frame() command?
>> Btw, comparing the graphics capabilities of GNUplot and R, it is
>> something like a three-wheel bicycle and a spaceship. Guess
>> which is which.
LH> =) I know that I will most likely spend a lot of time on just making
LH> the plots, but I atleast (for now =) ) think it could be fun to try.
if you make them with R, yes.
I wholeheartedly support Gabor's point:
I'd consider GNUplot to be clearly inferior to R -- just talking
about the graphics possibilties and the quality / thoughtfulness
in the high-level plotting.
If you have your data / objects / functions in R,
I'm very strongly convinced that using GNUplot for plotting is
``the wrong'' approach by almost all definitions of "wrong".
Martin Maechler, ETH Zurich