def test_sample_action():
n_actions = 2
len_list = 1
context = np.array([1.0, 1.0, 1.0, 1.0]).reshape(2, -1)
action = np.array([0, 1])
reward = np.array([1.0, 0.0])
position = np.array([0, 0])
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.fit(context=context,
action=action,
reward=reward,
position=position)
with pytest.raises(ValueError):
invalid_type_context = [1.0, 2.0]
learner.sample_action(context=invalid_type_context)
with pytest.raises(ValueError):
invalid_ndim_context = np.array([1.0, 2.0, 3.0, 4.0])
learner.sample_action(context=invalid_ndim_context)
context = np.array([1.0, 1.0, 1.0, 1.0]).reshape(2, -1)
n_rounds = context.shape[0]
sampled_action = learner.sample_action(context=context)
assert sampled_action.shape[0] == n_rounds
assert sampled_action.shape[1] == n_actions
assert sampled_action.shape[2] == len_list
def test_opl_predict():
n_actions = 2
len_list = 1
# shape error
with pytest.raises(ValueError):
context = np.array([1.0, 1.0])
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.predict(context=context)
# shape consistency of action_dist
# n_rounds is 5, dim_context is 2
context = np.array([1.0, 1.0, 1.0, 1.0]).reshape(2, -1)
action = np.array([0, 1])
reward = np.array([1.0, 0.0])
position = np.array([0, 0])
learner = IPWLearner(n_actions=2, len_list=1)
learner.fit(context=context,
action=action,
reward=reward,
position=position)
context_test = np.array([i for i in range(10)]).reshape(5, 2)
action_dist = learner.predict(context=context_test)
assert action_dist.shape[0] == 5
assert action_dist.shape[1] == n_actions
assert action_dist.shape[2] == len_list
def test_ipw_learner_predict():
n_actions = 2
len_list = 1
# shape error
desc = "`context` must be 2D array"
with pytest.raises(ValueError, match=f"{desc}*"):
context = np.array([1.0, 1.0])
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.predict(context=context)
# shape consistency of action_dist
# n_rounds is 5, dim_context is 2
context = np.array([1.0, 1.0, 1.0, 1.0]).reshape(2, -1)
action = np.array([0, 1])
reward = np.array([1.0, 0.0])
position = np.array([0, 0])
learner = IPWLearner(n_actions=2, len_list=1)
learner.fit(context=context,
action=action,
reward=reward,
position=position)
context_test = np.array([i for i in range(10)]).reshape(5, 2)
action_dist = learner.predict(context=context_test)
assert np.allclose(action_dist.sum(1),
np.ones_like((context_test.shape[0], len_list)))
assert action_dist.shape[0] == 5
assert action_dist.shape[1] == n_actions
assert action_dist.shape[2] == len_list
def test_create_train_data_for_opl():
context = np.array([1.0, 1.0]).reshape(1, -1)
learner = IPWLearner(n_actions=2)
action = np.array([0])
reward = np.array([1.0])
pscore = np.array([0.5])
X, sample_weight, y = learner._create_train_data_for_opl(
context=context, action=action, reward=reward, pscore=pscore
)
assert np.allclose(X, np.array([1.0, 1.0]).reshape(1, -1))
assert np.allclose(sample_weight, np.array([2.0]))
assert np.allclose(y, np.array([0]))
def test_ipw_learner_init_base_classifier_list():
# base classifier
len_list = 2
learner1 = IPWLearner(n_actions=2, len_list=len_list)
assert isinstance(learner1.base_classifier, LogisticRegression)
for i in range(len_list):
assert isinstance(learner1.base_classifier_list[i], LogisticRegression)
from sklearn.naive_bayes import GaussianNB
learner2 = IPWLearner(n_actions=2,
len_list=len_list,
base_classifier=GaussianNB())
assert isinstance(learner2.base_classifier, GaussianNB)
for i in range(len_list):
assert isinstance(learner2.base_classifier_list[i], GaussianNB)
def test_ipw_learner_init():
# base classifier
len_list = 2
learner1 = IPWLearner(n_actions=2, len_list=len_list)
assert isinstance(learner1.base_classifier, LogisticRegression)
for i in range(len_list):
assert isinstance(learner1.base_classifier_list[i], LogisticRegression)
with pytest.raises(ValueError):
from sklearn.linear_model import LinearRegression
IPWLearner(n_actions=2, base_classifier=LinearRegression())
from sklearn.naive_bayes import GaussianNB
learner2 = IPWLearner(n_actions=2, len_list=len_list, base_classifier=GaussianNB())
assert isinstance(learner2.base_classifier, GaussianNB)
for i in range(len_list):
assert isinstance(learner2.base_classifier_list[i], GaussianNB)
def test_ipw_learner_init_using_invalid_inputs(
n_actions,
len_list,
base_classifier,
description,
):
with pytest.raises(ValueError, match=f"{description}*"):
_ = IPWLearner(
n_actions=n_actions,
len_list=len_list,
base_classifier=base_classifier,
)
def test_ipw_learner_init_using_valid_inputs(
n_actions,
len_list,
base_classifier,
description,
):
ipw_learner = IPWLearner(
n_actions=n_actions,
len_list=len_list,
base_classifier=base_classifier,
)
# policy_type
assert ipw_learner.policy_type == PolicyType.OFFLINE
def test_base_opl_init():
# n_actions
with pytest.raises(ValueError):
IPWLearner(n_actions=1)
with pytest.raises(ValueError):
IPWLearner(n_actions="3")
# len_list
with pytest.raises(ValueError):
IPWLearner(n_actions=2, len_list=0)
with pytest.raises(ValueError):
IPWLearner(n_actions=2, len_list="3")
# policy_type
assert IPWLearner(n_actions=2).policy_type == "offline"
# invalid relationship between n_actions and len_list
with pytest.raises(ValueError):
IPWLearner(n_actions=5, len_list=10)
with pytest.raises(ValueError):
IPWLearner(n_actions=2, len_list=3)
def test_opl_fit():
context = np.array([1.0, 1.0, 1.0, 1.0]).reshape(2, -1)
action = np.array([0, 1])
reward = np.array([1.0, 0.0])
position = np.array([0, 0])
learner = IPWLearner(n_actions=2, len_list=1)
learner.fit(context=context,
action=action,
reward=reward,
position=position)
# inconsistency with the shape
with pytest.raises(AssertionError):
learner = IPWLearner(n_actions=2, len_list=2)
variant_context = np.array([1.0, 1.0, 1.0, 1.0])
learner.fit(context=variant_context,
action=action,
reward=reward,
position=position)
# len_list > 2, but position is not set
with pytest.raises(ValueError):
learner = IPWLearner(n_actions=2, len_list=2)
learner.fit(context=context, action=action, reward=reward)
def test_ipw_learner_fit():
n_rounds = 1000
dim_context = 5
n_actions = 3
len_list = 2
context = np.ones((n_rounds, dim_context))
action = np.random.choice(np.arange(len_list, dtype=int), size=n_rounds)
reward = np.random.choice(np.arange(2), size=n_rounds)
position = np.random.choice(np.arange(len_list, dtype=int), size=n_rounds)
# inconsistency with the shape
desc = "Expected `context.shape[0]"
with pytest.raises(ValueError, match=f"{desc}*"):
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
variant_context = np.random.normal(size=(n_rounds + 1, n_actions))
learner.fit(
context=variant_context,
action=action,
reward=reward,
position=position,
)
# len_list > 2, but position is not set
desc = "When `self.len_list > 1"
with pytest.raises(ValueError, match=f"{desc}*"):
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.fit(context=context, action=action, reward=reward)
# position must be non-negative
desc = "`position` elements must be non-negative integers"
with pytest.raises(ValueError, match=f"{desc}*"):
negative_position = position - 1
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.fit(context=context,
action=action,
reward=reward,
position=negative_position)
# IPWLearner cannot handle negative rewards
desc = "A negative value is found in"
with pytest.raises(ValueError, match=f"{desc}*"):
negative_reward = reward - 1.0
learner = IPWLearner(n_actions=n_actions, len_list=len_list)
learner.fit(context=context,
action=action,
reward=negative_reward,
position=position)