How to plot a hanging rootogram in python?

Question

Inspired by this question, how do you make the same kind of plot in python? This plot aims at having a nice visual representation of how your distribution is off of the expected distribution. It hangs the bars of your histogram to the expected distribution line, so the difference to the expected value is read between the bottom of the bar and the x-axis, instead of between the top of the bar and the expected distribution curve.

I could not find any built in function.

Answer 1

The idea is to just move each bar of the histogram plot where the top of the bar is at the expected value:

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.mlab as mlab

fig, ax = plt.subplots(1, 2)
mu = 10
sig = 0.3
my_data = np.random.normal(mu, sig, 200)
x = np.linspace(9, 11, 100)

# I plot the data twice, one for the histogram only for comparison,
# and one for the rootogram.
# The trick will be to modify the histogram to make it hang to
# the expected distribution curve:

for a in ax:
    a.hist(my_data, normed=True)
    a.plot(x, mlab.normpdf(x, mu, sig))
    a.set_ylim(-0.2)
    a.set_xlim(9, 11)
    a.hlines(0, 9, 11, linestyle="--")

for rectangle in ax[1].patches:

    # expected value in the middle of the bar
    exp = mlab.normpdf(rectangle.get_x() + rectangle.get_width()/2., mu, sig)

    # difference to the expected value
    diff = exp - rectangle.get_height()
    rectangle.set_y(diff)

    ax[1].plot(rectangle.get_x() + rectangle.get_width()/2., exp, "ro")

ax[0].set_title("histogram")
ax[1].set_title("hanging rootogram")
plt.tight_layout()

Which gives:

HTH