Python numpy random numbers probability

Question

Python 3.6.1 :: Anaconda custom (64-bit)

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

print (np.__version__)
1.12.1
print (mtptlb.__version__)
2.0.2

%matplotlib inline
a=np.random.uniform(1,100,1000000)
b=range(1,101)
plt.hist(a)

Why does the Y axis show 100000 ? np.random.uniform(1,100, 1000000 ) has the value 1000000, so shouldn't it show 1000000 on y axis?

Answer 1

By default matplotlib.pyplot.hist uses 10 bins. So all your 1 million values are distributed into 10 bins. For a perfect uniform distribution you would expect that you have 100k occurrences (1 million divided by 10) in each bin.

You can change the number of bins, ie

a=np.random.uniform(1, 100, 1000000)
plt.hist(a, bins=100)

Here it's divided into 100 bins and because it's a uniform distribution all bins are roughly at 10000.

Or just one bin if you want a count of 1 000 000:

a=np.random.uniform(1, 100, 1000000)
plt.hist(a, bins=1)