Group Bar Chart with Seaborn/Matplotlib

Question

My goal is to create a grouped bar chart like the one below, using a pandas DataFrame that is grouped by two variables "Alpha" and "Beta."

xl2 = xl.groupby(['Alpha','Beta']).median()

When I tried this, a KeyError was thrown on 'Alpha'

import seaborn as sns
sns.barplot(x=['Alpha', 'Beta'], y=xl2['Gamma'])

My hope was to pass in a list of x values to index on ('Alpha' and 'Beta'), and graph the associated 'Gamma." The documentation for the seaborn.barplot function doesn't provide any group bar chart examples.

Thanks for your help!

Are you looking for something like [this](http://matplotlib.org/examples/api/barchart_demo.html), or perhaps trying to use the `hue` variable in the documentation you've referenced? — Praveen, Apr 14 '16 at 18:17

score 4 · Answer 1 · answered May 27 '16 at 20:42

You can use ggplot for this

from ggplot import *
import pandas as pd
import numpy as np

df = pd.DataFrame({
    "x": np.random.choice(range(2001, 2008), 250),
    "w": np.random.uniform(50, 400, 250),
    "cat": np.random.choice(["A", "B", "C", "D", "E"], 250)
})

print ggplot(df, aes(x='x', weight='w', fill='cat')) + geom_bar() + theme_bw()

score 0 · Answer 2 · edited May 23 '17 at 12:17

is that what you want?

In [167]: df
Out[167]:
    a  b  c
0   2  2  1
1   3  3  1
2   2  2  1
3   2  3  0
4   3  2  2
5   3  3  2
6   1  2  2
7   1  2  2
8   0  2  3
9   3  2  3
10  2  2  0
11  2  1  2
12  2  1  0
13  1  2  1
14  0  2  3
15  0  3  3
16  3  1  2
17  0  1  1
18  0  2  2
19  0  1  0

In [168]: plot = df.groupby(['a','b']).mean()

In [169]: plot
Out[169]:
            c
a b
0 1  0.500000
  2  2.666667
  3  3.000000
1 2  1.666667
2 1  1.000000
  2  0.666667
  3  0.000000
3 1  2.000000
  2  2.500000
  3  1.500000

In [170]: sns.barplot(x=plot.index, y=plot.c)

PS if you need something different, please provide a sample data set and expected grouped resulting DF (both in text/dict/JSON/CSV form)

PPS you may also want to check this answer

Almost. The index is composed of two items. I want one item (Beta) to appear alongside the graph as a legend. Currently, the x-label is formatted in the following way ('Alpha', Beta) — NumenorForLife, Apr 15 '16 at 17:42

score -1 · Answer 3 · answered Jul 06 '17 at 02:59

Altair can be helpful in such cases. Here is the plot produced by the following code.

Imports

import pandas as pd
import numpy as np
from altair import *

Generating dataset

np.random.seed(0)
df = pd.DataFrame({
    "x": np.random.choice(range(0, 5), 250),
    "w": np.random.uniform(50, 400, 250),
    "cat": np.random.choice(["A", "B", "C", "D", "E"], 250)
})

Plotting

Chart(df).mark_bar().encode(x=X('cat', axis=False),  
                            y=Y('median(w)', axis=Axis(grid=False)),
                            color='cat',
                            column=Column('x', axis=Axis(axisWidth=1.0, offset=-8.0, orient='bottom'),scale=Scale(padding=30.0)),
                        ).configure_facet_cell( strokeWidth=0.0).configure_cell(width=200, height=200)

The key things in the altair code are:

X values are categories ('cat' in the df)
Color is by category
Y values are by median of the variable
Different columns represent different years

Group Bar Chart with Seaborn/Matplotlib

3 Answers3

Imports

Generating dataset

Plotting

Linked