def GradientDatastream():
dataset = datastream.datasets()["gradient"]
augmenter = datastream.augmenter()
return Datastream.merge([
Datastream(dataset.subset(lambda df: df["class_name"] == class_name))
for class_name in settings.CLASS_NAMES
]).map(lambda example: example.augment(augmenter))
def GradientDatastream():
dataset = evaluate_datastreams()['gradient'].dataset
augmenter_ = augmenter()
return (
Datastream.merge([
Datastream(dataset.subset(
lambda df: df['class_name'] == class_name
))
for class_name in settings.CLASS_NAMES
])
.map(lambda example: example.augmented(augmenter_))
)
def evaluate_datastreams():
evaluate_datasets = problem.evaluate_datasets()
return {
split_name: Datastream(dataset)
for split_name, dataset in dict(
gradient=evaluate_datasets['train'],
compare=evaluate_datasets['compare'],
).items()
}
def evaluate_datastreams():
evaluate_datasets = problem.evaluate_datasets()
evaluate_datasets['train'] = evaluate_datasets['train'].split(
key_column='index',
proportions=dict(gradient=0.8, early_stopping=0.2),
stratify_column='class_name',
filepath='{{cookiecutter.package_name}}/splits/early_stopping.json',
)
return {
split_name: Datastream(dataset)
for split_name, dataset in dict(
gradient=evaluate_datasets['train']['gradient'],
early_stopping=evaluate_datasets['train']['early_stopping'],
compare=evaluate_datasets['compare'],
).items()
}
def evaluate_datastreams():
return {
split_name: Datastream(dataset)
for split_name, dataset in datastream.datasets().items()
}
class ResourceApi:
datastream = Datastream()
sensor = Sensor()
reading = Reading()
statio = Station()
# Load the backend as specified in configuration
if getattr(settings, 'DATASTREAM_BACKEND', None) is not None:
backend = settings.DATASTREAM_BACKEND
if isinstance(backend, basestring):
i = backend.rfind('.')
module, attr = backend[:i], backend[i + 1:]
try:
mod = importlib.import_module(module)
except ImportError, e:
raise exceptions.ImproperlyConfigured(
'Error importing datastream backend %s: "%s"' % (module, e))
try:
cls = getattr(mod, attr)
except AttributeError:
raise exceptions.ImproperlyConfigured(
'Module "%s" does not define a "%s" datastream backend' %
(module, attr))
backend = cls(**getattr(settings, 'DATASTREAM_BACKEND_SETTINGS', {}))
def callback(stream_id, granularity, datapoint):
signals.new_datapoint.send(sender=datastream,
stream_id=stream_id,
granularity=granularity,
datapoint=datapoint)
datastream = Datastream(backend, callback)
def init_datastream(datastream_backend, datastream_backend_settings):
backend = datastream_backend
if isinstance(backend, basestring):
i = backend.rfind('.')
module, attr = backend[:i], backend[i + 1:]
try:
mod = importlib.import_module(module)
except ImportError, exception:
raise exceptions.ImproperlyConfigured(
"Error importing datastream backend %s: %s" %
(module, exception))
try:
cls = getattr(mod, attr)
except AttributeError:
raise exceptions.ImproperlyConfigured(
"Module '%s' does not define a '%s' datastream backend" %
(module, attr))
backend = cls(**datastream_backend_settings)
return Datastream(backend)
# Load the backend as specified in configuration
if getattr(settings, 'DATASTREAM_BACKEND', None) is not None:
datastream = init_datastream(
settings.DATASTREAM_BACKEND,
getattr(settings, 'DATASTREAM_BACKEND_SETTINGS', {}))