EasyCalculusca-247

Fisher Information for Bernoulli

Problem

For a Bernoulli random variable with success probability p, the Fisher information is I(p) = 1 / (p * (1 - p)).

Given p with 0 < p < 1, return I(p). The information is symmetric about p = 0.5 and diverges at the boundaries.

Hint

The Fisher information is the expected squared score, which is largest near fair coins.

case 1

in: [0.5]

expected: 4

case 2

in: [0.25]

expected: 5.333333333333333

case 3

in: [0.1]

expected: 11.111111111111109

case 4

in: [0.9]

expected: 11.111111111111112

Starter codePython

def fisher_bernoulli(p):
    # Your code here
    pass

4 test cases · run your code against them in the editor