Show median and quantiles on Seaborn pairplot (Python)-CodePudding

I am making a corner plot using Seaborn. I would like to display lines on each diagonal histogram showing the median value and quantiles. Example shown below.

I usually do this using the Python package 'corner', which is straightforward. I want to use Seaborn just because it has better aesthetics.

The seaborn plot was made using this code:

import seaborn as sns    

df = pd.DataFrame(samples_new, columns = ['r1', 'r2', 'r3'])

cornerplot = sns.pairplot(df, corner=True, kind='kde',diag_kind="hist", diag_kws={'color':'darkslateblue', 'alpha':1, 'bins':10}, plot_kws={'color':'darkslateblue', 's':10, 'alpha':0.8, 'fill':False})

CodePudding user response：

Seaborn provides test data sets that come in handy to explain something you want to change to the default behavior. That way, you don't need to generate your own test data, nor to supply your own data that can be complicated and/or sensitive.

To update the subplots in the diagonal, there is g.map_diag(...) which will call a given function for each individual column. It gets 3 parameters: the data used for the x-axis, a label and a color.

Here is an example to add vertical lines for the main quantiles, and change the title. You can add more calculations for further customizations.

import matplotlib.pyplot as plt
import seaborn as sns

def update_diag_func(data, label, color):
    for val in data.quantile([.25, .5, .75]):
        plt.axvline(val, ls=':', color=color)
    plt.title(data.name, color=color)

iris = sns.load_dataset('iris')
g = sns.pairplot(iris, corner=True, diag_kws={'kde': True})
g.map_diag(update_diag_func)
g.fig.subplots_adjust(top=0.97) # provide some space for the titles
plt.show()

CodePudding user response：

Seaborn is built ontop of matplotlib so you can try this:

import seaborn as sns    
from matplotlib import pyplot as plt

df = pd.DataFrame(samples_new, columns = ['r1', 'r2', 'r3'])

cornerplot = sns.pairplot(df, corner=True, kind='kde',diag_kind="hist", diag_kws={'color':'darkslateblue', 'alpha':1, 'bins':10}, plot_kws={'color':'darkslateblue', 's':10, 'alpha':0.8, 'fill':False})

plt.text(300, 250, "An annotation")
plt.show()