def set_attrs(self,
target=None,
features=None,
estimator=None,
prediction=None,
predictions_name=None,
actual=None,
column_subset=None,
filters=None,
fill_missing=None,
discard_incomplete=False,
categorical_indicators=False):
if prediction is not None:
if predictions_name is None:
raise ValueError("If you provide a prediction feature, you "
"must also specify a _unique_ 'predictions_name'")
if isinstance(target, BaseFeature) or target is None:
self.target = target
else:
self.target = Feature(target)
if isinstance(prediction, BaseFeature) or prediction is None:
self.prediction = prediction
else:
self.prediction = Feature(prediction)
self.predictions_name = predictions_name
if actual is None:
actual = self.target
self.actual = (actual if isinstance(actual, BaseFeature)
else Feature(actual))
self.filters = filters if filters else []
if discard_incomplete:
self.filters.append(filter_incomplete)
if features:
self.features = ([f if isinstance(f, BaseFeature) else Feature(f)
for f in features])
if categorical_indicators:
self.features = pre_transform_features(self.features,
AsFactorIndicators,
only_if_categorical=True)
if fill_missing is not None:
self.features = pre_transform_features(self.features,
FillMissing,
fill_value=missing)
else:
self.features = None
# Wrap estimator to return probabilities in the case of a classifier
self.estimator = wrap_sklearn_like_estimator(estimator)
self.column_subset = column_subset
def set_attrs(self,
target=None,
features=None,
estimator=None,
evaluation_transformation=None,
predictions_name=None,
evaluation_target=None,
column_subset=None,
filters=None,
fill_missing=None,
discard_incomplete=False,
categorical_indicators=False):
if isinstance(target, BaseFeature) or target is None:
self.target = target
else:
self.target = Feature(target)
# Alternative predictions
if isinstance(evaluation_transformation, BaseFeature) or evaluation_transformation is None:
self.evaluation_transformation = evaluation_transformation
else:
self.evaluation_transformation = Feature(evaluation_transformation)
self.predictions_name = predictions_name
if self.predictions_name is None:
self.predictions_name = self.DEFAULT_PREDICTIONS_NAME
if evaluation_target is not None:
evaluation_target = (evaluation_target if isinstance(evaluation_target, BaseFeature)
else Feature(evaluation_target))
self.evaluation_target = evaluation_target
if (self.evaluation_target is not None) ^ (self.evaluation_transformation is not None):
raise ValueError("You must specify both or neither of\
evaluation_target and evaluation_transformation")
# Transformations
self.filters = filters if filters else []
if discard_incomplete:
self.filters.append(filter_incomplete)
self.fill_missing = fill_missing
self.categorical_indicators = categorical_indicators
self.discard_incomplete = discard_incomplete
# Features
if features:
self.features = ([f if isinstance(f, BaseFeature) else Feature(f)
for f in features])
if categorical_indicators:
self.features = pre_transform_features(self.features,
AsFactorIndicators,
only_if_categorical=True)
if fill_missing is not None:
self.features = pre_transform_features(self.features,
FillMissing,
fill_value=fill_missing)
else:
self.features = None
# Wrap estimator to return probabilities in the case of a classifier
self.estimator = wrap_sklearn_like_estimator(estimator)
self.column_subset = column_subset
def set_attrs(self,
target=None,
features=None,
estimator=None,
evaluation_transformation=None,
predictions_name=None,
evaluation_target=None,
column_subset=None,
filters=None,
fill_missing=None,
discard_incomplete=False,
categorical_indicators=False):
if isinstance(target, BaseFeature) or target is None:
self.target = target
else:
self.target = Feature(target)
# Alternative predictions
if isinstance(evaluation_transformation,
BaseFeature) or evaluation_transformation is None:
self.evaluation_transformation = evaluation_transformation
else:
self.evaluation_transformation = Feature(evaluation_transformation)
self.predictions_name = predictions_name
if self.predictions_name is None:
self.predictions_name = self.DEFAULT_PREDICTIONS_NAME
if evaluation_target is not None:
evaluation_target = (evaluation_target if isinstance(
evaluation_target, BaseFeature) else
Feature(evaluation_target))
self.evaluation_target = evaluation_target
if (self.evaluation_target
is not None) ^ (self.evaluation_transformation is not None):
raise ValueError("You must specify both or neither of\
evaluation_target and evaluation_transformation")
# Transformations
self.filters = filters if filters else []
if discard_incomplete:
self.filters.append(filter_incomplete)
self.fill_missing = fill_missing
self.categorical_indicators = categorical_indicators
self.discard_incomplete = discard_incomplete
# Features
if features:
self.features = ([
f if isinstance(f, BaseFeature) else Feature(f)
for f in features
])
if categorical_indicators:
self.features = pre_transform_features(
self.features,
AsFactorIndicators,
only_if_categorical=True)
if fill_missing is not None:
self.features = pre_transform_features(self.features,
FillMissing,
fill_value=fill_missing)
else:
self.features = None
# Wrap estimator to return probabilities in the case of a classifier
self.estimator = wrap_sklearn_like_estimator(estimator)
self.column_subset = column_subset
