Bar Chart with multiple labels

Question

The following code only shows the main category ['one', 'two', 'three', 'four', 'five', 'six'] as the x axis labels. Is there a way show subcategory ['A', 'B', 'C', 'D'] as secondary x axis labels?

df = pd.DataFrame(np.random.rand(6, 4),
                 index=['one', 'two', 'three', 'four', 'five', 'six'],
                 columns=pd.Index(['A', 'B', 'C', 'D'], 
                 name='Genus')).round(2)


df.plot(kind='bar',figsize=(10,4))

Tutor

I can think of two options: 1. Create an ad-hoc secondary x-axis below the primary one (see this) ; 2. You start with df.unstack().plot.bar() and you change the figure attributes afterwards. – Jacquejacquelin

Answer 1

Here a possible solution (I had quite a lot of fun!):

df = pd.DataFrame(np.random.rand(6, 4),
                 index=['one', 'two', 'three', 'four', 'five', 'six'],
                 columns=pd.Index(['A', 'B', 'C', 'D'],
                 name='Genus')).round(2)

ax = df.plot(kind='bar',figsize=(10,4), rot = 0)

# "Activate" minor ticks
ax.minorticks_on()

# Get location of the center of each rectangle
rects_locs = map(lambda x: x.get_x() +x.get_width()/2., ax.patches)
# Set minor ticks there
ax.set_xticks(rects_locs, minor = True)


# Labels for the rectangles
new_ticks = reduce(lambda x, y: x + y, map(lambda x: [x] * df.shape[0], df.columns.tolist()))
# Set the labels
from matplotlib import ticker
ax.xaxis.set_minor_formatter(ticker.FixedFormatter(new_ticks))  #add the custom ticks

# Move the category label further from x-axis
ax.tick_params(axis='x', which='major', pad=15)

# Remove minor ticks where not necessary
ax.tick_params(axis='x',which='both', top='off')
ax.tick_params(axis='y',which='both', left='off', right = 'off')

Here's what I get:

Inviolable

Nice trick to use minor and major tick labels for this; it won't work, however, if there's an odd number of bars per group as you can't have both minor and major labels for the same tick mark. So in this case you'll need to add Texts for the group labels. – Gelatinate

Answer 2

Here is a solution. You can get the positions of the bars and set some minor xticklabels accordingly.

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

df = pd.DataFrame(np.random.rand(6, 4),
                 index=['one', 'two', 'three', 'four', 'five', 'six'],
                 columns=pd.Index(['A', 'B', 'C', 'D'], 
                 name='Genus')).round(2)


df.plot(kind='bar',figsize=(10,4))

ax = plt.gca()
pos = []
for bar in ax.patches:
    pos.append(bar.get_x()+bar.get_width()/2.)


ax.set_xticks(pos,minor=True)
lab = []
for i in range(len(pos)):
    l = df.columns.values[i//len(df.index.values)]
    lab.append(l)

ax.set_xticklabels(lab,minor=True)
ax.tick_params(axis='x', which='major', pad=15, size=0)
plt.setp(ax.get_xticklabels(), rotation=0)

plt.show()

Ivar

Nice trick to use minor and major tick labels for this; it won't work, however, if there's an odd number of bars per group as you can't have both minor and major labels for the same tick mark. So in this case you'll need to add Texts for the group labels. – Gelatinate

Answer 3

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

def subcategorybar(X, vals,als, width=0.8):
    n = len(vals)
    _X = np.arange(len(X))
    plt.figure(figsize=(14,9))
    for i in range(n):
        plt.bar(_X - width/2. + i/float(n)*width, vals[i], 
                width=width/float(n), align="edge")
        for j in _X:
            plt.text([_X - width/2. + i/float(n)*width][0][j],vals[i][j]+0.01*vals[i] 
                     [j],str(als[i][j]))
    plt.xticks(_X, X)

### data
X = ['a','b','c','d','f']
A1 = [1,2,3,4,5]
A2= [1,7,6,7,8]
A3 = [3,5,6,8,9]
A4= [4,5,6,7,3]
A5 = [5,6,7,8,5]

##labels
A1_al = ['da','dd',5,6,3]
A2_al = np.random.random_integers(20,size=5)
A3_al = np.random.random_integers(20,size=5)
A4_al = np.random.random_integers(20,size=5)
A5_al = np.random.random_integers(20,size=5)

subcategorybar(X, [A1,A2,A3,A4],[A1_al,A2_al,A3_al,A4_al],width=0.8)

plt.show()

Hackamore