/ggplot

ggplot for python

Primary LanguagePythonBSD 2-Clause "Simplified" LicenseBSD-2-Clause

{ggplot} from Yhat

read more on our blog

from ggplot import *

ggplot(aes(x='date', y='beef'), data=meat) + \
    geom_point(color='lightblue') + \
    geom_line(alpha=0.25) + \
    stat_smooth(span=.05, color='black') + \
    ggtitle("Beef: It's What's for Dinner") + \
    xlab("Date") + \
    ylab("Head of Cattle Slaughtered")

What is it?

Yes, it's another port of ggplot2. One of the biggest reasons why I continue to reach for R instead of Python for data analysis is the lack of an easy to use, high level plotting package like ggplot2. I've tried other libraries like bokeh and d3py but what I really want is ggplot2.

ggplot is just that. It's an extremely un-pythonic package for doing exactly what ggplot2 does. The goal of the package is to mimic the ggplot2 API. This makes it super easy for people coming over from R to use, and prevents you from having to re-learn how to plot stuff.

Goals

  • same API as ggplot2 for R
  • never use matplotlib again
  • ability to use both American and British English spellings of aesthetics
  • tight integration with pandas
  • pip installable

Getting Started

Dependencies

I realize that these are not fun to install. My best luck has always been using brew if you're on a Mac or just using the binaries if you're on Windows. If you're using Linux then this should be relatively painless. You should be able to apt-get or yum all of these.

  • matplotlib
  • pandas
  • numpy
  • scipy
  • statsmodels

Installation

Ok the hard part is over. Installing ggplot is really easy. Just use pip! An item on the TODO is to add the matplotlibrc files to the pip installable (let me know if you'd like to help!).

# matplotlibrc from Huy Nguyen (http://www.huyng.com/posts/sane-color-scheme-for-matplotlib/)
$ curl https://github.com/yhat/ggplot/raw/master/matplotlibrc.zip > matplotlibrc.zip 
$ unzip matplotlibrc.zip -d ~/
# install ggplot using pip
$ pip install ggplot

Loading ggplot

# run an IPython shell (or don't)
$ ipython
In [1]: from ggplot import *

That's it! You're ready to go!

Examples

meat_lng = pd.melt(meat[['date', 'beef', 'pork', 'broilers']], id_vars='date')
ggplot(aes(x='date', y='value', colour='variable'), data=meat_lng) + \
    geom_point() + \
    stat_smooth(color='red')

####geom_point

from ggplot import *
ggplot(diamonds, aes('carat', 'price')) + \
    geom_point(alpha=1/20.) + \
    ylim(0, 20000)

####geom_hist

p = ggplot(aes(x='carat'), data=diamonds)
p + geom_hist() + ggtitle("Histogram of Diamond Carats") + labs("Carats", "Freq") 

####geom_density

ggplot(diamonds, aes(x='price', color='cut')) + \
    geom_density()

meat_lng = pd.melt(meat[['date', 'beef', 'broilers', 'pork']], id_vars=['date'])
p = ggplot(aes(x='value', colour='variable', fill=True, alpha=0.3), data=meat_lng)
p + geom_density()

####geom_bar

p = ggplot(mtcars, aes('factor(cyl)'))
p + geom_bar()

TODO

The list is long, but distinguished.TODO