Strange behavior in matplotlib (multiple) histograms

Question

I tried to plot two histograms on the same canvas, using the standard method:

    plt.figure(figsize=(8,6))
    plt.hist(samp_1_I, label=['Female'], alpha=0.5)
    plt.hist(samp_0_I, label=['Male'], alpha=0.5)
    plt.tick_params(axis='both', which='major', labelsize=18)
    plt.legend(fontsize=18)
    plt.show()

and the result is: If you notice, histogram 'bars' are not aligned towards the end (right most side). How to fix it? Has anyone seen this before?

The horizontal axis are rational numbers [0,1] and by default it is binned into 10 bin of 0.1. I don't understand why bar of two groups are not aligned as in the beginning.

Answer 1

If you want 10 bins with binwidth 0.1, you need to provide those bins in the call to plt.hist .

import matplotlib.pyplot as plt
import numpy as np

samp_1_I = np.random.rand(14)
samp_0_I = np.random.rand(17)

bins= np.arange(11)/10.

plt.figure(figsize=(8,6))
plt.hist(samp_1_I, bins=bins, label=['Female'], alpha=0.5)
plt.hist(samp_0_I, bins=bins, label=['Male'], alpha=0.5)
plt.tick_params(axis='both', which='major', labelsize=18)
plt.legend(fontsize=18)
plt.show()

---

Answering the question from the comments: Using normed=True you'll get the normalized frequency in the sense that the sum over each bin width times the the frequency is 1 .

sample = np.random.rand(14)
bins= np.arange(11)/10.
hist, _bins = np.histogram(sample, bins=bins, normed = True)
print hist
print np.sum ( hist * np.diff(bins) ) # This prints 1.0

This holds even if bin widths are different.