Python: boxplot dataframe with multi-index

Question

I am trying to create a boxplot, where on the x-axis I will refer to the two columns of the dataframe, while on y-axis I will show values of the 3rd column.

Let me refer to an example dataframe:

  Lvl1  Lvl2  value
0    A     1      1
1    A     2      2
2    A     1      3
3    B     2      4
4    B     1      5
5    B     2      6

Now, I want to have boxplots for the groups corresponding to Lvl1 and Lvl2. For example for group represented by (Lvl1 = A, Lvl2 = 1) boxplot would be calculated of values of {1,3}.

I know I can create a new column like Lvl0 which would be something like Lvl1 + Lvl2, but is there a way to create a boxplot without such operation?

On the following code:

import pandas as pd
import matplotlib.pyplot as plt

dataset = pd.DataFrame(
    {'Lvl1': ['A', 'A', 'A', 'B', 'B', 'B'], 'Lvl2': [1, 2, 1, 2, 1, 2], 'value': [1, 2, 3, 4, 5, 6]})
grouped = dataset.groupby(['Lvl1', 'Lvl2'])
grouped.boxplot()
plt.show()

I get an error:
KeyError: "None of [Index(['A', 1], dtype='object')] are in the [index]"

Thank you in advance!

Answer 1

Try to use seaborn for an easier solution. I think it was answered here: Grouping boxplots in seaborn when input is a DataFrame

With your data:

import seaborn as sns
import pandas as pd

data = pd.DataFrame({'lvl1': ['A', 'A', 'A', 'B', 'B', 'B'], 
                     'lvl2': [1, 2, 1, 2, 1, 2], 
                     'value': [1, 2, 3, 4, 5, 6]})

df_long = pd.melt(data, "lvl1", var_name="lvl2", value_name="result")

sns.boxplot(x="lvl1", hue="lvl2", y="result", data=df_long)

We get:

Answer 2

You can do it through seaborn. Following code works for me on your data:

    import pandas as pd
    import seaborn as sns

    dataset = pd.DataFrame(
    {'Lvl1': ['A', 'A', 'A', 'B', 'B', 'B'], 'Lvl2': [1, 2, 1, 2, 1, 2], 
    'value': [1, 2, 3, 4, 5, 6]})

    ax = sns.boxplot(x='Lvl1', y='value', hue="Lvl2",
             data=dataset)

Desired plot