def demo(output_file=None, instances=40000):
""" _test_prequential_bagging
This demo shows the evaluation process of a LeverageBagging classifier,
initialized with KNN classifiers.
Parameters
----------
output_file: string
The name of the csv output file
instances: int
The evaluation's max number of instances
"""
# Setup the File Stream
# opt = FileOption("FILE", "OPT_NAME", "../datasets/sea_big.csv", "CSV", False)
# stream = FileStream(opt, -1, 1)
stream = SEAGenerator(classification_function=2,
instance_seed=755437,
noise_percentage=0.0)
stream.prepare_for_use()
# Setup the classifier
#classifier = OzaBaggingAdwin(h=KNN(k=8, max_window_size=2000, leaf_size=30, categorical_list=None))
classifier = LeverageBagging(h=KNN(k=8, max_window_size=2000,
leaf_size=30),
ensemble_length=1)
# Setup the pipeline
pipe = Pipeline([('Classifier', classifier)])
# Setup the evaluator
eval = EvaluatePrequential(
pretrain_size=2000,
max_instances=instances,
batch_size=1,
n_wait=200,
max_time=1000,
output_file=output_file,
task_type='classification',
show_plot=True,
plot_options=['kappa', 'kappa_t', 'performance'])
# Evaluate
eval.eval(stream=stream, classifier=pipe)
def demo():
""" _test_streams
This demo tests if the streams are correctly generating samples.
:return:
"""
opt = FileOption('FILE', 'OPT_NAME', '../datasets/covtype.csv', 'csv',
False)
stream = FileStream(opt, -1, 1)
stream.prepare_for_use()
rbf_drift = RandomRBFGeneratorDrift(change_speed=41.00,
num_centroids=50,
model_seed=32523423,
instance_seed=5435,
num_classes=2,
num_att=10,
num_drift_centroids=50)
rbf_drift.prepare_for_use()
sea = SEAGenerator()
print('1 instance:\n')
X, y = stream.next_instance()
print(X)
print(y)
X, y = sea.next_instance()
print(X)
print(y)
print('\n\n10 instances:\n')
X, y = stream.next_instance(10)
print(X)
print(y)
X, y = sea.next_instance(10)
print(X)
print(y)
def demo():
""" _test_leverage_bagging
This demo tests the LeverageBagging classifier on a file stream, which gives
instances coming from a SEA generator.
The test computes the performance of the LeverageBagging classifier as well
as the time to create the structure and classify max_samples (2000 by default)
instances.
"""
logging.basicConfig(format='%(message)s', level=logging.INFO)
warnings.filterwarnings("ignore", ".*Passing 1d.*")
stream = SEAGenerator(1, noise_percentage=6.7)
stream.prepare_for_use()
clf = LeverageBagging(h=KNN(k=8, max_window_size=2000, leaf_size=30),
ensemble_length=1)
sample_count = 0
correctly_classified = 0
max_samples = 2000
train_size = 200
first = True
if train_size > 0:
X, y = stream.next_instance(train_size)
clf.partial_fit(X, y, classes=stream.get_classes())
first = False
logging.info('%s%%', 0.0)
while sample_count < max_samples:
if (sample_count + 1) % (max_samples / 20) == 0:
logging.info('%s%%',
str(((sample_count // (max_samples / 20) + 1) * 5)))
X, y = stream.next_instance(2)
my_pred = clf.predict(X)
if first:
clf.partial_fit(X, y, classes=stream.get_classes())
first = False
else:
clf.partial_fit(X, y)
if my_pred is not None:
if y[0] == my_pred[0]:
correctly_classified += 1
sample_count += 1
print(str(sample_count) + ' samples analyzed.')
print('My performance: ' + str(correctly_classified / sample_count))
print(clf.get_info())
def demo():
""" _test_oza_bagging
This demo tests the OzaBagging classifier using KNNAdwin classifiers,
on samples given by a SEAGenerator.
The test computes the performance of the OzaBagging classifier as well
as the time to create the structure and classify max_samples (5000 by
default) instances.
"""
logging.basicConfig(format='%(message)s', level=logging.INFO)
warnings.filterwarnings("ignore", ".*Passing 1d.*")
stream = SEAGenerator(1, noise_percentage=6.7)
stream.prepare_for_use()
#print(stream.get_targets())
clf = OzaBagging(h=KNNAdwin(k=8, max_window_size=2000, leaf_size=30),
ensemble_length=2)
sample_count = 0
correctly_classified = 0
max_samples = 5000
train_size = 8
first = True
if train_size > 0:
X, y = stream.next_sample(train_size)
clf.partial_fit(X, y, classes=stream.get_targets())
first = False
while sample_count < max_samples:
if sample_count % (max_samples / 20) == 0:
logging.info('%s%%', str((sample_count // (max_samples / 20) * 5)))
X, y = stream.next_sample()
my_pred = clf.predict(X)
if first:
clf.partial_fit(X, y, classes=stream.get_targets())
first = False
else:
clf.partial_fit(X, y)
if my_pred is not None:
if y[0] == my_pred[0]:
correctly_classified += 1
sample_count += 1
print(str(sample_count) + ' samples analyzed.')
print('My performance: ' + str(correctly_classified / sample_count))
def test_sea_generator(test_path):
stream = SEAGenerator(classification_function=2, instance_seed=112, balance_classes=False, noise_percentage=0.28)
stream.prepare_for_use()
assert stream.estimated_remaining_instances() == -1
expected_header = ['att_num_0', 'att_num_1', 'att_num_2']
assert stream.get_attributes_header() == expected_header
expected_classes = [0, 1]
assert stream.get_classes() == expected_classes
assert stream.get_classes_header() == ['class']
assert stream.get_num_attributes() == 3
assert stream.get_num_nominal_attributes() == 0
assert stream.get_num_numerical_attributes() == 3
assert stream.get_num_targets() == 2
assert stream.get_num_values_per_nominal_attribute() == 0
assert stream.get_plot_name() == 'SEA Generator - 2 class labels'
assert stream.has_more_instances() is True
assert stream.is_restartable() is True
# Load test data corresponding to first 10 instances
test_file = os.path.join(test_path, 'sea_stream.npz')
data = np.load(test_file)
X_expected = data['X']
y_expected = data['y']
X, y = stream.next_instance()
assert np.alltrue(X[0] == X_expected[0])
assert np.alltrue(y[0] == y_expected[0])
X, y = stream.get_last_instance()
assert np.alltrue(X[0] == X_expected[0])
assert np.alltrue(y[0] == y_expected[0])
stream.restart()
X, y = stream.next_instance(10)
assert np.alltrue(X == X_expected)
assert np.alltrue(y == y_expected)
def test_sea_generator(test_path):
stream = SEAGenerator(classification_function=2,
random_state=112,
balance_classes=False,
noise_percentage=0.28)
stream.prepare_for_use()
assert stream.n_remaining_samples() == -1
expected_names = ['att_num_0', 'att_num_1', 'att_num_2']
assert stream.feature_names == expected_names
expected_targets = [0, 1]
assert stream.target_values == expected_targets
assert stream.target_names == ['target_0']
assert stream.n_features == 3
assert stream.n_cat_features == 0
assert stream.n_num_features == 3
assert stream.n_targets == 1
assert stream.get_data_info(
) == 'SEA Generator - 1 targets, 2 classes, 3 features'
assert stream.has_more_samples() is True
assert stream.is_restartable() is True
# Load test data corresponding to first 10 instances
test_file = os.path.join(test_path, 'sea_stream.npz')
data = np.load(test_file)
X_expected = data['X']
y_expected = data['y']
X, y = stream.next_sample()
assert np.alltrue(X[0] == X_expected[0])
assert np.alltrue(y[0] == y_expected[0])
X, y = stream.last_sample()
assert np.alltrue(X[0] == X_expected[0])
assert np.alltrue(y[0] == y_expected[0])
stream.restart()
X, y = stream.next_sample(10)
assert np.alltrue(X == X_expected)
assert np.alltrue(y == y_expected)
assert stream.n_targets == np.array(y).ndim
assert stream.n_features == X.shape[1]