python - Function to create grouped bar plot

17
votes

The goal here is to create a grouped bar plot, not subplots like the image below

Is there a simple way to create a grouped bar plot in Python? Right now I get separate bar plots, instead of separate bars on one plot.

df = pd.DataFrame([['g1','c1',10],['g1','c2',12],['g1','c3',13],['g2','c1',8],['g2','c2',10],['g2','c3',12]],columns=['group','column','val'])

%matplotlib inline
df.groupby(['group']).plot(kind='bar')

enter image description here

2
pythonpython-3.xpandasmatplotlibbar-chart
I guess this indeed not a duplicate of the mentionned question. However, it might still have been asked before (feels like a standard thing to do), so if anyone finds a dup for this, we might close again. Until then see my answer below. - ImportanceOfBeingErnest
For incoming readers, this is not a 'bar chart' (en.wikipedia.org/wiki/Bar_chart), but a mere grouped scatter plot using 'bar' instead of 'dot' to show data point, don't be misled. - Jason Goal
@JasonGoal, This is indeed a bar plot (the data is showing relationships among categories). Although I will admit to using a poor naming convention in my example. Please see this part of the definition you linked above: "One axis of the chart shows the specific categories being compared, and the other axis represents a measured value. Some bar graphs present bars clustered in groups of more than one, showing the values of more than one measured variable." - Rilcon42

2 Answers

31
votes

Pandas will show grouped bars by columns. Entries in each row but different columns will constitute a group in the resulting plot. Hence you need to "reshape" your dataframe to have the "group" as columns. In this case you can pivot like

df.pivot("column", "group", "val")

producing 

group   g1  g2
column        
c1      10   8
c2      12  10
c3      13  12

Plotting this will result in a grouped bar chart.

import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame([['g1','c1',10],['g1','c2',12],['g1','c3',13],['g2','c1',8],
                   ['g2','c2',10],['g2','c3',12]],columns=['group','column','val'])

df.pivot("column", "group", "val").plot(kind='bar')

plt.show()

enter image description here

3
votes

You can simply do this using the code given below:

import pandas as pd
import matplotlib.pyplot as plt

positive_values = [20, 17.5, 40]
negative_values = [15, 8, 70]
index = ['Precision', 'Recall', 'f1-score',]
df = pd.DataFrame({'Positive Values': positive_values,
                    'Negative Values': negative_values}, index=index)
ax = df.plot.bar(rot=0, color={"Positive Values": "green", "Negative Values": "red"})

Output:

Output