Combine two dataframe boxplots in a twinx figure

Question

I want to display two Pandas dataframes within one figure as boxplots. As each of the two dataframes has different value range, I would like to have them combined in a twinx figure.

Reduced to the minimum, I have tried the following:

import pandas as pd import numpy as np import matplotlib.pyplot as plt df1 = pd.DataFrame(np.random.randint(0,100,size=(100, 4)), columns=list('ABCD')) df2 = pd.DataFrame(np.random.randint(100,200,size=(100, 2)), columns=list('EF')) fig, ax1 = plt.subplots() ax2 = ax1.twinx() df1.boxplot(ax=ax1) df2.boxplot(ax=ax2) plt.show()

The result is expectedly not what it should look like (there should be 6 boxes on the plot, actually!)

How can I manage to have the boxplots next to each other? I tried to set some dummy scatter points on ax1 and ax2, but this did not really help.

Can you make an example with random data that we can actually run? — Chiel
– Chiel, Commented Apr 8, 2016 at 8:32

Chiel · Accepted Answer · 2016-04-08 09:32:19Z

The best solution is to concatenate the data frames for plotting and to use a mask. In the creation of the mask, we use the dfs == dfs | dfs.isnull() to create a full matrix with True and then we query on all column names that are not 'E' or 'F'. This gives a 2D matrix that allows you to only plot the first four boxes, as the last two two are masked (so their ticks do appear at the bottom). With the inverse mask ~mask you plot the last two on their own axis and mask the first four.

import pandas as pd import numpy as np import matplotlib.pyplot as plt df1 = pd.DataFrame(np.random.randint( 0,100,size=(100, 4)), columns=list('ABCD')) df2 = pd.DataFrame(np.random.randint(100,200,size=(100, 2)), columns=list('EF' )) dfs = pd.concat([df1, df2]) mask = ((dfs == dfs) | dfs.isnull()) & (dfs.columns != 'E') & (dfs.columns != 'F') fig, ax1 = plt.subplots() dfs[mask].boxplot() ax2 = ax1.twinx() dfs[~mask].boxplot() plt.show()

Collectives™ on Stack Overflow

Combine two dataframe boxplots in a twinx figure

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related