def _make_sampler(self,
append=True,
random_state=None,
**overwrite_kwargs):
"""Make and configure a copy of the `base_sampler_` attribute.
Warning: This method should be used to properly instantiate new
sub-samplers.
"""
sampler = clone(self.base_sampler_)
if hasattr(self, 'sampler_kwargs_'):
sampler.set_params(**self.sampler_kwargs_)
# Arguments passed to _make_sampler function have higher priority,
# they will overwrite the self.sampler_kwargs_
sampler.set_params(**overwrite_kwargs)
if random_state is not None:
_set_random_states(sampler, random_state)
if append:
self.samplers_.append(sampler)
return sampler
def _clone_estimator(base_estimator, random_state=None):
estimator = clone(base_estimator)
if random_state is not None:
_set_random_states(estimator, random_state)
return estimator
def _clone_estimator(base_estimator, random_state=None, idx=0):
estimator = clone(base_estimator)
if random_state is not None:
random_state = 255 if random_state == 0 else random_state
_set_random_states(estimator, random_state * 37 * (idx + 1))
return estimator
def _make_estimator(self, append=True, random_state=None):
"""Make and configure a copy of the `estimator_` attribute.
Warning: This method should be used to properly instantiate new
sub-estimators.
"""
estimator = clone(self.estimator_)
estimator.set_params(**{p: getattr(self, p) for p in self.estimator_params})
if random_state is not None:
_set_random_states(estimator, random_state)
if append:
self.estimators_.append(estimator)
return estimator
def _make_sampler_estimator(self, random_state=None):
"""Make and configure a copy of the `base_estimator_` attribute.
Warning: This method should be used to properly instantiate new
sub-estimators.
"""
estimator = clone(self.base_estimator_)
estimator.set_params(
**{p: getattr(self, p)
for p in self.estimator_params})
sampler = clone(self.base_sampler_)
if random_state is not None:
_set_random_states(estimator, random_state)
_set_random_states(sampler, random_state)
return estimator, sampler
def _validate_estimator(self, random_state):
"""Private function to create the classifier"""
if (self.estimator is not None
and isinstance(self.estimator, ClassifierMixin)
and hasattr(self.estimator, "predict_proba")):
self.estimator_ = clone(self.estimator)
_set_random_states(self.estimator_, random_state)
elif self.estimator is None:
self.estimator_ = RandomForestClassifier(
n_estimators=100,
random_state=self.random_state,
n_jobs=self.n_jobs,
)
else:
raise ValueError(
f"Invalid parameter `estimator`. Got {type(self.estimator)}.")
def _make_sampler_estimator(self, append=True, random_state=None):
"""Make and configure a copy of the `base_estimator_` attribute.
Warning: This method should be used to properly instantiate new
sub-estimators.
"""
estimator = clone(self.base_estimator_)
estimator.set_params(**{p: getattr(self, p) for p in self.estimator_params})
sampler = clone(self.base_sampler_)
if random_state is not None:
_set_random_states(estimator, random_state)
_set_random_states(sampler, random_state)
if append:
self.estimators_.append(estimator)
self.samplers_.append(sampler)
self.pipelines_.append(
make_pipeline(deepcopy(sampler), deepcopy(estimator))
)
return estimator, sampler
def test_set_random_states():
# Linear Discriminant Analysis doesn't have random state: smoke test
_set_random_states(LinearDiscriminantAnalysis(), random_state=17)
clf1 = Perceptron(random_state=None)
assert clf1.random_state is None
# check random_state is None still sets
_set_random_states(clf1, None)
assert isinstance(clf1.random_state, int)
# check random_state fixes results in consistent initialisation
_set_random_states(clf1, 3)
assert isinstance(clf1.random_state, int)
clf2 = Perceptron(random_state=None)
_set_random_states(clf2, 3)
assert clf1.random_state == clf2.random_state
# nested random_state
def make_steps():
return [('sel', SelectFromModel(Perceptron(random_state=None))),
('clf', Perceptron(random_state=None))]
est1 = Pipeline(make_steps())
_set_random_states(est1, 3)
assert isinstance(est1.steps[0][1].estimator.random_state, int)
assert isinstance(est1.steps[1][1].random_state, int)
assert (est1.get_params()['sel__estimator__random_state'] !=
est1.get_params()['clf__random_state'])
# ensure multiple random_state parameters are invariant to get_params()
# iteration order
class AlphaParamPipeline(Pipeline):
def get_params(self, *args, **kwargs):
params = Pipeline.get_params(self, *args, **kwargs).items()
return OrderedDict(sorted(params))
class RevParamPipeline(Pipeline):
def get_params(self, *args, **kwargs):
params = Pipeline.get_params(self, *args, **kwargs).items()
return OrderedDict(sorted(params, reverse=True))
for cls in [AlphaParamPipeline, RevParamPipeline]:
est2 = cls(make_steps())
_set_random_states(est2, 3)
assert (est1.get_params()['sel__estimator__random_state'] ==
est2.get_params()['sel__estimator__random_state'])
assert (est1.get_params()['clf__random_state'] ==
est2.get_params()['clf__random_state'])
