matplotlib grouped bars in a bar graph

Question

I'm playing with matplotlib and pandas to create a bar graph. I have a dataframe of world cup teams and here is what df.head() shows

I want to create a bar graph, where each bar is a unique team , and each team in a group are graphed close together, with unique groups between a noticeable distance apart. Here is what I currently have:

caps_per_team_per_group = df.groupby(['group', 'team']).Caps.sum()
caps_per_team_per_group.plot(kind='bar', figsize=(15,5))

and for visual purposes, I want it to look like this:

Answer 1

The most straightforward way to do this is to reset the index of your groupby data frame and use seaborn's factorplot.

import numpy as np
import pandas as pd
import seaborn as sns

# Setting up
group_A = ['Egypt', 'Russia', 'Uruguay', 'Saudi Arabia']
group_B = ['Spain', 'Portugal', 'Iran', 'Morocco']
group_C = ['France', 'Denmark', 'Australia', 'Peru']
team = group_A + group_B + group_C
group =['A', 'A', 'A', 'A', 'B', 'B', 'B', 'B', 'C', 'C', 'C', 'C']
caps = np.random.randint(5, 50, 12)
caps_per_team_per_group = pd.DataFrame({ 'Caps': caps}, index=pd.MultiIndex.from_tuples(list(zip(group, team)), names=['Group', 'Team']))

# Reset index and plot
caps_per_team_per_group.reset_index(inplace=True)
g = sns.factorplot(x="Group", y="Caps", hue='Team', data=caps_per_team_per_group, kind="bar", palette="muted")

n [47]: caps_per_team_per_group
Out[47]: 
                    Caps
Group Team              
A     Egypt           43
      Russia          39
      Uruguay         20
      Saudi Arabia    49
B     Spain            6
      Portugal         8
      Iran            32
      Morocco         11
C     France          14
      Denmark         41
      Australia       29
      Peru             8