如何创建分组条 plot

Question

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.

import pandas as pd

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'])

  group column  val
0    g1     c1   10
1    g1     c2   12
2    g1     c3   13
3    g2     c1    8
4    g2     c2   10
5    g2     c3   12
    

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

Answer 1

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()

Answer 2

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:

Answer 3

Plotly express is one of the best visualisation packages I've recently used. It allows you to generate visualisations without needing to perform massive data transformations.

# initial dataframe
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.head()

   group column val
0   g1   c1     10
1   g1   c2     12
2   g1   c3     13
3   g2   c1      8
4   g2   c2     10
5   g2   c3     12

No need to transform data, directly use plotly express:

import plotly.express as px
fig = px.bar(df, x="column", y="val",
             color='group', barmode='group',text="val",
             height=400)
fig.show()

Answer 4

• Given a dataframe of long data, as shown in the OP, an implementation that does not require transforming the dataframe is to useseaborn.barplot with the hue parameter.
• seaborn is a high-level API for matplotlib
• Tested with seaborn 0.11.1 and matplotlib 3.4.2

import pandas as pd
import seaborn as sns

# the sample dataframe from the OP
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'])

# plot with seaborn barplot
sns.barplot(data=df, x='column', y='val', hue='group')