C[omp]ute

From: andrew cooke <andrew@...>

Date: Tue, 24 Jul 2012 20:17:46 -0400

I just finished reading "ggplot2" by Hadley Wickham (published by Springer
which means it's horribly expensive, even in paperback).

It's a good book and a good system.  If you've not heard of ggplot2, it's a
plotting package for R.  If you've not heard of R it's an environment /
programming language for statistics.

ggplot2 is structured around a "grammar of graphics" which is, as far as I can
see, a domain model for plots.  That means that the plots, and the commands
you use to generate them, are structured around quite high-level concets like
"aesthetics", "geometries" and "scales".

So you need to get used to that.

But once you do, there are some big wins.  You only have to change small
things to get radically different plots - an example in the book goes from
an area plot (think mountain range) to bubbles (floating circls scaled to
reflect the data).

And while some of this is obvious from the online docs (and multiple
websites), the book goes into more detail, including themes, which allow you
to change the visual style to match a particular publication, for example.

If anyone wanted a medium to large size project that would be a pretty much
guaranteed win, implementing this in Python for use with scipy etc would be a
great idea...

Andrew

PS One downside - my copy, at least (the cheapest I could find - paperback and
remaindered), was printed completely in black and white, which makes the
illustrations showing colour use pointless.