def __init__(self,
steps,
*,
memory=None,
verbose=False,
output_name=None,
enforce_float32=True,
runtime='python',
options=None,
white_op=None,
black_op=None,
final_types=None,
op_version=None):
self.output_name = output_name
self.enforce_float32 = enforce_float32
self.runtime = runtime
self.options = options
self.white_op = white_op
self.white_op = white_op
self.black_op = black_op
self.final_types = final_types
self.op_version = op_version
# The constructor calls _validate_step and it checks the value
# of black_op.
Pipeline.__init__(self, steps, memory=memory, verbose=verbose)
def __init__(self, steps):
Pipeline.__init__(self, steps)
if not isinstance(self._dataframe_mapper, DataFrameMapper):
raise TypeError(
"First step of a DataFramePipeline must be a DataFrameMapper, "
"'%s' (type %s) is not." %
(self._dataframe_mapper, type(self._dataframe_mapper))
)
def __init__(self, nQuantile=None):
self.nQuantile = Separator_num.nQUANTILE if nQuantile is None else nQuantile
Pipeline.__init__(
self,
[
('geometry', Separator_num.Selector()),
('quantiled',
QuantileTransformer(n_quantiles=self.nQuantile, copy=False)
) #use in-place scaling
])
def __init__(self, k,
model_type='random_forest',
kmeans_type='squishy',
n_forests=192,
n_trees=1,
n_features_to_predict=0.5,
max_depth=5, #should be 2 for boosting
learning_rate=0.6,
using_weights=True,
weight_extent=1, # 2 for boosting
max_iter=60,
n_attempts=10,
weight_adjustment=0,
eig_extent=0,
n_jobs=1):
slc1 = SLCluster(n_forests,
model_type=model_type,
n_trees=n_trees,
n_features_to_predict=n_features_to_predict,
max_depth=max_depth,
outputting_weights=False,
weight_extent=weight_extent,
learning_rate=learning_rate,
n_jobs=n_jobs)
slc2 = SLCluster(n_forests,
model_type=model_type,
n_trees=n_trees,
n_features_to_predict=n_features_to_predict,
max_depth=max_depth,
outputting_weights=using_weights,
weight_extent=weight_extent,
learning_rate=learning_rate,
n_jobs=n_jobs)
ew = EigenvectorWeighting(extent=eig_extent)
if kmeans_type == 'normal':
jk = JKMeans(k,
max_iter=max_iter,
n_attempts=n_attempts,
accepting_weights=using_weights,
weight_adjustment=weight_adjustment,
n_jobs=n_jobs)
else:
jk = SquishyJKMeans(k,
max_iter=max_iter,
n_attempts=n_attempts,
accepting_weights=using_weights,
weight_adjustment=weight_adjustment,
n_jobs=n_jobs)
if eig_extent == 0:
Pipeline.__init__(self,[('slc1', slc1), ('slc2', slc2), ('jkmeans', jk)])
else:
Pipeline.__init__(self,[('slc1', slc1), ('slc2', slc2), ('ew', ew), ('jkmeans', jk)])
def __init__(
self,
k,
model_type='random_forest',
kmedoids_type='normal',
n_forests=150,
n_trees=1,
n_features_to_predict=0.5,
max_depth=5, #should be 2 if model_type is boosting
learning_rate=0.6,
using_weights=False,
using_pca=False,
weight_extent=1, # 2 if model_type is boosting
max_iter=60,
n_attempts=10,
weight_adjustment=0,
eig_extent=0,
n_jobs=1):
rft = RFTransform(n_forests,
model_type=model_type,
n_trees=n_trees,
n_features_to_predict=n_features_to_predict,
max_depth=max_depth,
outputting_weights=using_weights,
using_pca=using_pca,
weight_extent=weight_extent,
learning_rate=learning_rate,
n_jobs=n_jobs)
ew = EigenvectorWeighting(extent=eig_extent)
if kmedoids_type == 'normal':
jk = JKMedoids(k,
max_iter=max_iter,
n_attempts=n_attempts,
accepting_weights=using_weights,
weight_adjustment=weight_adjustment,
n_jobs=n_jobs)
else:
jk = SquishyJKMedoids(k,
max_iter=max_iter,
n_attempts=n_attempts,
accepting_weights=using_weights,
weight_adjustment=weight_adjustment,
n_jobs=n_jobs)
if eig_extent == 0 or not using_weights:
Pipeline.__init__(self, [('rft', rft), ('jkmeans', jk)])
else:
Pipeline.__init__(self, [('rft', rft), ('ew', ew),
('jkmeans', jk)])
def __init__(self, steps, *, memory=None, verbose=False,
output_name=None, enforce_float32=True,
runtime='python', options=None,
white_op=None, black_op=None, final_types=None,
op_version=None):
Pipeline.__init__(
self, steps, memory=memory, verbose=verbose)
self.output_name = output_name
self.enforce_float32 = enforce_float32
self.runtime = runtime
self.options = options
self.white_op = white_op
self.white_op = white_op
self.black_op = black_op
self.final_types = final_types
self.op_version = op_version
def __init__(self, steps=None, frc_mdl=None, gen_mdl=None, sel_mdl=None):
# if steps is None and frc_mdl is None:
# raise ValueError("Steps are not defined in Pipeline model")
if steps is None:
steps = [('gen', gen_mdl), ('sel', sel_mdl), ('frc', frc_mdl)]
named_steps = {k: v for k, v in steps}
if named_steps['frc'] is None:
frc_mdl = IdentityModel(name="Identity")
if named_steps['sel'] is None:
sel_mdl = sel_class.FeatureSelection(name="No feature selection",
on=False)
if named_steps['gen'] is None:
gen_mdl = gnt_class.FeatureGeneration(name="No feature generation")
steps = [('gen', gen_mdl), ('sel', sel_mdl), ('frc', frc_mdl)]
Pipeline.__init__(self, steps)
self.name = "_".join(
[str(frc_mdl.name),
str(gen_mdl.name),
str(sel_mdl.name)])
def __init__(self, steps, na_val=np.nan):
Pipeline.__init__(self, steps)
self.na_val = na_val
def __init__(self, steps, memory=None, verbose=False, op_version=None):
Pipeline.__init__(self, steps=steps, memory=memory, verbose=verbose)
OnnxSubGraphOperatorMixin.__init__(self)
self.op_version = op_version
def __init__(self, steps):
Pipeline.__init__(self, steps)
def __init__(self, steps, cache_name=None, verbose=False):
Pipeline.__init__(self, steps, memory=None, verbose=verbose)
if cache_name is None:
cache_name = "Pipeline%d" % id(self)
self.cache_name = cache_name
def __init__(self, steps):
Pipeline.__init__(self, steps)
def __init__(self, op_version=None):
Pipeline.__init__(self)
OnnxSubGraphOperatorMixin.__init__(self)
self.op_version = op_version
