def __init__(self):
samples = []
classes = []
for i in range(10):
for j in range(10):
samples.append([10 * i + j])
classes.append(i)
Environment.__init__(self, np.array(samples), np.array(classes))
class TestEnvironment:
"""
Functions tested in TestEnvironment:
- get_random_sample
- get_random_sample_index
- get_sample
- get_samples
Functions not tested:
- __init__
- standard_distance
- get_all
- get_all_classes
- get_all_samples
- get_class
- set_distance
"""
def __init__(self):
irises = datasets.load_iris()
self.env = Environment(irises.data, irises.target)
def get_random(self):
_, sample, _ = self.env.get_random_sample()
return sample
def get_index(self):
index, _, _ = self.env.get_random_sample()
return index
def get_sample(self, index):
return self.env.get_sample(index)
def test_get_random_sample_multiple_sampling_gives_more_than_one_sample(self):
sample = self.get_random()
different = False
for _ in range(1000):
new_sample = self.get_random()
if (new_sample == sample).all():
different = True
break
assert different
def test_get_random_sample_index_multiple_sampling_gives_more_than_one_sample(self):
sample = self.get_index()
different = False
for _ in range(1000):
new_sample = self.get_index()
if new_sample == sample:
different = True
break
assert different
def test_get_sample_the_same_results_like_normal_indexes_x_100(self):
irises = datasets.load_iris()
for _ in range(100):
index = self.get_index()
assert (irises.data[index] == self.get_sample(index)).all()
def test_get_samples_full_returned_array_identical_with_original(self):
original = self.env.get_all_samples()
copy = self.env.get_samples(range(len(original)))
assert (original == copy).all()
def __init__(self):
irises = datasets.load_iris()
self.env = Environment(irises.data, irises.target)
def __init__(self):
self.agent = None
irises = datasets.load_iris()
self.environment = Environment(irises.data, irises.target)
class TestAgent:
"""
Functions tested in TestAgent:
- classify
Functions not tested:
- __init__
- add_sample
- decrease_weight_word_category
- forget
- increase_weights_sample_category
- increase_weight_word_category
- learn
- update_fitness_measure
- decrease_weights_for_other_categories
- decrease_weights_for_other_words
- get_category_class
- get_id
- get_fitness_measure
- get_words
- set_fitness
"""
def __init__(self):
self.agent = None
irises = datasets.load_iris()
self.environment = Environment(irises.data, irises.target)
def add(self, sample_index, category=None, environment=None):
environment = environment or self.environment
self.agent.add_sample(sample_index, environment, category)
def classify(self, sample=None):
if sample is None:
sample = self.environment.get_random_sample()
return self.agent.predict(sample)
def setup(self):
self.agent = SteelsAgent()
def test_classify(self):
sample_index, _, _ = self.environment.get_random_sample()
sample = self.environment.get_sample(sample_index)
# Classify with no samples returns None
assert_equals(self.classify(sample), None)
# Classify with one sample returns this sample category
self.add(1, 2)
assert_equals(self.classify(sample), 2)
# Classify before fitting (learning) returns None
self.add(51, 1)
assert_equals(self.classify(sample), None)
# Classify after fitting (learning) doesn't return None
self.agent.learn()
assert self.classify(sample) is not None
def test_learning_cycle(self):
self.add(1, 1)
self.add(51, 2)
self.add(101, 3)
self.agent.learn()
for _ in range(10):
assert self.classify() is not None
def test_single_classification_game_iterations(self):
agent = self.agent
env = self.environment
agent.set_fitness("DF", DummyFitness())
for _ in range(100):
agent.learn()
sample_index = env.get_random_sample_index()
sample = env.get_sample(sample_index)
category = agent.predict(sample)
if category is None:
self.add(sample_index)
value = 0
elif env.get_class(sample_index) == agent.get_category_class(category):
agent.increase_weights_sample_category(sample_index, env, category)
value = 1
elif agent.get_fitness_measure("DF") > 0.95:
agent.increase_weights_sample_category(sample_index, env, category)
value = 0
else:
self.add(sample_index)
value = 0
self.agent.update_fitness_measure("DF", value)
self.agent.forget()
