Seaborn Barplot With Regression Line

Is there a way to add a regression line to a barplot in seaborn where the x axis contains pandas.Timestamps? For example, overlay a trendline in this bar plot below. Am looking fo

Solution 1:

Seaborn barplots are categorical plots. Categorical plots cannot directly be used for regression, because the numeric values would not fit. The usual matplotlib bar plots however use numeric data.

An option is to plot a matplotlib barplot and seaborn regplot in the same graph.

import numpy as np; np.random.seed(1)
import seaborn.apionly as sns
import matplotlib.pyplot as plt

x = np.linspace(5,9,13)
y = np.cumsum(np.random.rand(len(x)))

fig, ax = plt.subplots()

ax.bar(x,y, width=0.1, color="lightblue", zorder=0)
sns.regplot(x=x, y=y, ax=ax)
ax.set_ylim(0, None)
plt.show()

Since seaborn's barplot uses the integers from 0 to number of bars as indizes, one can also use those indizes for a regression plot on top of seaborn bar plot.

import numpy as np
import seaborn.apionly as sns
import matplotlib.pyplot as plt
import pandas

sns.set(style="white", context="talk")
a = pandas.DataFrame.from_dict({'Attendees': {pandas.Timestamp('2016-12-01'): 10,
  pandas.Timestamp('2017-01-01'): 12,
  pandas.Timestamp('2017-02-01'): 15,
  pandas.Timestamp('2017-03-01'): 16,
  pandas.Timestamp('2017-04-01'): 20}})
ax = sns.barplot(data=a, x=a.index, y=a.Attendees, color='lightblue' )
# put bars in background:
for c in ax.patches:
    c.set_zorder(0)
# plot regplot with numbers 0,..,len(a) as x value
sns.regplot(x=np.arange(0,len(a)), y=a.Attendees, ax=ax)
sns.despine(offset=10, trim=False)
ax.set_ylabel("")
ax.set_xticklabels(['Dec', 'Jan','Feb','Mar','Apr'])
plt.show()