Epsilon-Greedy Choice

Problem

Choose an arm with epsilon-greedy. Call random.seed(seed); with probability epsilon pick a uniformly random arm, otherwise pick the arm with the largest value (ties keep the smallest index).

Empty values return -1.

Hint

Epsilon controls how often the policy explores instead of exploiting.

Test cases

case 1

in: [[1,3,2],0.1,42]

expected: 1

case 2

in: [[1,3,2],1,7]

expected: 0

case 3

in: [[1,3,2],0.5,0]

expected: 1

case 4

in: [[5,5],0,7]

expected: 0

Starter codePython

import random
def epsilon_greedy_choice(values, epsilon, seed):
    # Your code here
    pass

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