Pointwise Mutual Information

Problem

Compute the pointwise mutual information in bits:

PMI = log2(P(x, y) / (P(x) * P(y)))

The joint and marginal probabilities are passed as separate arguments.

Hint

PMI is positive when the events co-occur more than chance would predict.

case 1

in: [0.25,0.5,0.5]

expected: 0

case 2

in: [0.5,0.5,0.5]

expected: 1

case 3

in: [0.2,0.5,0.5]

expected: -0.3219280948873623

case 4

in: [0.125,0.5,0.5]

expected: -1

Starter codePython

import math
def pointwise_mi(pxy, px, py):
    # Your code here
    pass

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