Pandas/matplotlib bar chart with colors defined by column

Question

I am trying to make a bar chart in python, and color bars by a data frame column, this is exceedingly easy in R ggplot2, so I really do who not understand why it is so hard in matplotlib/pandas, I would like to understand how to do it and hopefully the logic as I assume it can't be hard after all

Here is an example of what I want. I did in ggplot - all I want is to define the color using attributes in the data frame, the attribute can be the color string i.e. 'k','r' .. or a feature of the data say male/female, etc.

this is an example of code I tried that works to generate the bars

import matplotlib.pyplot as plt
import pandas as pd
data = pd.DataFrame({'name' : ['a','b','c'],'value' : [1,2,3], 
                     'c' : ['b','r','b']})
data.plot('name','value',kind='bar')

but I want the color defined by column 'c'

but when I add color='c' then it does not work

data.plot('name','value',color='c',kind='bar',color='c')

I tried using plot.scatter as well and it did not work either

---- Update

I did some more work and this works more or less, I have yet to figure out how to properly align the labels, I still do to like why I have to have a numeric x axis, when it is really categorical (again ggplot2 deals with this on its own) - but ar least this can be done in code

fig, ax = plt.subplots()
ax.bar(range(len(data)),data['value'],width=.5,color=data['c'])
ax.set_xticks(range(len(data)))
ax.set_xticklabels( (data['name'] ))

Thanks as always

Solution

Try this:

data.plot('name','value',color=['r', 'g', 'b'],kind='bar')

You can give any combination of colors as a list to the color argument. If there are more bars than number of colors, the colors will just recycle.

I can also strongly recommend the excellent brewer2mpl library. It provides aesthetically pleasing color choices. The code looks like this:

import brewer2mpl
bmap = brewer2mpl.get_map('Set2','qualitative',3,reverse=True)
colors = bmap.mpl_colors
data.plot('name','value',color=colors,kind='bar')

resulting in:

You can get brewer2mpl from here: https://pypi.python.org/pypi/brewer2mpl/1.4