def set_warm_start(self, warm_start):
"""
Set the status of `warm_start` of the Classifier.
Parameters
----------
warm_start : bool
Determines warm starting to allow training to pick
up from previous training sessions.
"""
TimeSeriesClassifier.set_warm_start(self, warm_start)
for e in (self.layer_1, self.layer_2):
e.set_warm_start(warm_start)
def set_verbose(self, verbose):
"""
Set the verbosity of the Classifier.
Subsidiary Classifiers have their verbosity
suppressed by one compared to this Classifier.
Parameters
----------
verbose : int, default=0
Determines the verbosity of cross-validation.
Higher verbose levels result in more output logged.
"""
TimeSeriesClassifier.set_verbose(self, verbose)
for e in (self.layer_1, self.layer_2):
e.set_verbose(verbose - 1)
def set_warm_start(self, warm_start):
"""
Set the status of `warm_start` of the Classifier.
Parameters
----------
warm_start : bool
Determines warm starting to allow training to pick
up from previous training sessions.
"""
TimeSeriesClassifier.set_warm_start(self, warm_start)
estimators = self.estimators_ if hasattr(
self, 'estimators_') else self.base_estimators
for e in estimators:
e.set_warm_start(warm_start)
def set_verbose(self, verbose):
"""
Set the verbosity of the Classifier.
Subsidiary Classifiers have their verbosity
suppressed by one compared to this Classifier.
Parameters
----------
verbose : int, default=0
Determines the verbosity of cross-validation.
Higher verbose levels result in more output logged.
"""
TimeSeriesClassifier.set_verbose(self, verbose)
estimators = self.estimators_ if hasattr(
self, 'estimators_') else self.base_estimators
for e in estimators:
e.set_verbose(verbose - 1)
def set_random_state(self, random_state):
"""
Set the RandomState of the Classifier.
Subsidiary Classifiers set their RandomState
based on this RandomState with a seed selected
from 0 to 2**16.
Parameters
----------
random_state : None or int or RandomState, default=None
Initial seed for the RandomState. If `random_state` is None,
return the RandomState singleton. If `random_state` is an int,
return a RandomState with the seed set to the int.
If `random_state` is a RandomState, return that RandomState.
"""
TimeSeriesClassifier.set_random_state(self, random_state)
for e in (self.layer_1, self.layer_2):
e.set_random_state(self.random_state.randint(0, 2**16))
def __init__(self,
estimators=default_estimators,
warm_start=False,
metric='accuracy',
random_state=None,
verbose=0):
TimeSeriesClassifier.__init__(self,
warm_start=warm_start,
metric=metric,
random_state=random_state,
verbose=verbose)
for e in estimators:
check_estimator(e)
self.n = len(estimators)
self.base_estimators = estimators
self.set_warm_start(warm_start)
self.set_metric(metric)
self.set_verbose(verbose)
self.set_random_state(random_state)
def set_metric(self, metric):
"""
Set the metric of the Classifier.
Parameters
----------
metric : Metric, None, str
Metric to look up. Must be one of:
- 'accuracy' : Accuracy.
- 'precision' : Precision.
- 'recall' : Recall.
- 'f-score' : F1-Score.
- 'roc-auc' : ROC-AUC.
- Metric : A custom implementation.
- None : Return None.
Custom Metrics must implement `score` which
by default should return a single float value.
"""
TimeSeriesClassifier.set_metric(self, metric)
for e in (self.layer_1, self.layer_2):
e.set_metric(metric)
def _is_fitted(self):
"""
Returns if the Classifier has been trained and is
ready to predict new data.
Returns
-------
fitted : bool
True if the Classifier is fitted, False otherwise.
"""
attributes = ["estimators_", "n_classes_", "n_features_"]
return TimeSeriesClassifier._is_fitted(self, attributes=attributes)
def __init__(self,
layer_1=TimeSeriesEnsemble(),
layer_2=TimeSeriesEnsemble(),
warm_start=False,
metric='accuracy',
random_state=None,
verbose=0):
TimeSeriesClassifier.__init__(self,
warm_start=warm_start,
metric=metric,
random_state=random_state,
verbose=verbose)
if not issubclass(type(layer_1), TimeSeriesClassifier):
raise ValueError("Layer 1 must be a TimeSeriesClassifier")
if not issubclass(type(layer_2), TimeSeriesClassifier):
raise ValueError("Layer 2 must be a TimeSeriesClassifier")
self.layer_1 = layer_1
self.layer_2 = layer_2
self.set_warm_start(warm_start)
self.set_metric(metric)
self.set_verbose(verbose)
self.set_random_state(random_state)