def test_egreedy_select_action_exploration():
trial_num = 50
policy = EpsilonGreedy(n_actions=2, epsilon=1.0)
policy.action_counts = np.array([3, 3])
policy.reward_counts = np.array([3, 0])
selected_action = [policy.select_action() for _ in range(trial_num)]
assert 0 < sum(selected_action)[0] < trial_num
def test_egreedy_select_action_exploitation():
trial_num = 50
policy = EpsilonGreedy(n_actions=2, epsilon=0.0)
policy.action_counts = np.array([3, 3])
policy.reward_counts = np.array([3, 0])
for _ in range(trial_num):
assert policy.select_action()[0] == 0