MAP with Beta Prior

Problem

For a Bernoulli parameter with a Beta(alpha, beta) prior, the maximum a posteriori estimate after successes successes in trials trials is (alpha + successes - 1) / (alpha + beta + trials - 2). Return 0.5 when the denominator is zero and 0.0 for invalid arguments.

Hint

The Beta MAP behaves like the MLE after subtracting one pseudo-count from each prior parameter.

Test cases

case 1

in: [2,3,5,10]

expected: 0.46153846153846156

case 2

in: [1,1,3,10]

expected: 0.3

case 3

in: [1,1,0,0]

expected: 0.5

case 4

in: [3,2,0,4]

expected: 0.2857142857142857

case 5

in: [0,1,1,1]

expected: 0

Starter codePython

def beta_map_estimate(alpha, beta, successes, trials):
    # Your code here
    pass

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