def get_space_with_dynamic(self):
space = HyperSpace()
with space.as_default():
input1 = HyperInput()
input2 = HyperInput()
id1 = Identity(p1=Choice([1, 2]), p2=Int(1, 100))
id2 = Identity(p3=Real(0, 1))
id3 = Identity(p4=Dynamic(lambda p5: p5 * 3, p5=Choice([2, 4, 8])))
def new_module(m):
dm1 = Identity(dp1=Choice(['a', 'b']))
return dm1, dm1
id4 = ConnectionSpace(dynamic_fn=new_module, p6=Choice(['f', 'g']))
id5 = Identity()
id6 = Identity()
id7 = Identity()
id1([input1, id7])
id2([input1, input2])
id3([input2, id1, id2])
id4([input1, id2])
id5(input2)
id6(id3)
id7([id4, id5])
return space
def test_basic_ops(self):
with HyperSpace().as_default() as space:
id1 = HyperInput()
id2 = HyperInput()
id3 = Identity()
id3([id1, id2])
id4 = Identity()
id4(id3)
id5 = Identity()
id5(id3)
graph_inputs = space.get_inputs()
id3_inputs = space.get_inputs(id3)
id4_inputs = space.get_inputs(id4)
assert graph_inputs == [id1, id2]
assert id3_inputs == [id1, id2]
assert id4_inputs == [id3]
graph_outputs = space.get_outputs()
id1_outputs = space.get_outputs(id1)
id3_outputs = space.get_outputs(id3)
assert graph_outputs == [id4, id5]
assert id1_outputs == [id3]
assert id3_outputs == [id4, id5]
space.disconnect(id3, id5)
id3_outputs = space.get_outputs(id3)
assert id3_outputs == [id4]
assert len(space.modules) == 5
assert len(space.edges) == 3
def get_space_numeric_pipeline_complex():
space = HyperSpace()
with space.as_default():
input = HyperInput(name='input1')
p1 = numeric_pipeline_complex()(input)
p3 = DataFrameMapper(input_df=True, df_out=True)([p1]) # passthrough
est = LightGBMEstimator(task='binary', fit_kwargs={})(p3)
space.set_inputs(input)
return space
def get_space_num_cat_pipeline(default=False):
space = HyperSpace()
with space.as_default():
input = HyperInput(name='input1')
p1 = numeric_pipeline_simple()(input)
p2 = categorical_pipeline_simple()(input)
p3 = DataFrameMapper(default=default, input_df=True,
df_out=True)([p1, p2]) # passthrough
est = LightGBMEstimator(task='binary', fit_kwargs={})(p3)
space.set_inputs(input)
return space
def get_space(self):
space = HyperSpace()
with space.as_default():
input1 = HyperInput()
input2 = HyperInput()
id1 = Identity(p1=Choice([1, 2]), p2=Int(1, 100))
id2 = Identity(p3=Real(0, 1))
id3 = Identity(p4=Dynamic(lambda p5: p5 * 3, p5=Choice([2, 4, 8])))
id4 = Identity()
id5 = Identity()
id6 = Identity()
id7 = Identity()
id1([input1, id7])
id2([input1, input2])
id3([input2, id1, id2])
id4([input1, id2])
id5(input2)
id6(id3)
id7([id4, id5])
return space
def func_space(func):
space = HyperSpace()
with space.as_default():
params = {
name: copy.copy(v)
for name, v in zip(func.__code__.co_varnames, func.__defaults__)
if isinstance(v, ParameterSpace)
}
for _, v in params.items():
v.attach_to_space(space, v.name)
input = HyperInput()
id1 = Identity(**params)(input)
space.set_outputs([id1])
return space
def get_space_p_in_p():
space = HyperSpace()
with space.as_default():
p1 = Pipeline(
[SimpleImputer(name='imputer1'),
StandardScaler(name='scaler1')],
name='p1')
p2 = Pipeline(
[SimpleImputer(name='imputer2'),
StandardScaler(name='scaler2')],
name='p2')
input = HyperInput(name='input1')
p3 = Pipeline([p1, p2], name='p3')(input)
space.set_inputs(input)
return space
def get_space_num_cat_pipeline_complex(dataframe_mapper_default=False,
lightgbm_fit_kwargs={},
xgb_fit_kwargs={},
catboost_fit_kwargs={}):
space = HyperSpace()
with space.as_default():
input = HyperInput(name='input1')
p1 = numeric_pipeline_complex()(input)
p2 = categorical_pipeline_complex()(input)
# p2 = categorical_pipeline_simple()(input)
p3 = DataFrameMapper(default=dataframe_mapper_default,
input_df=True,
df_out=True,
df_out_dtype_transforms=[(column_object, 'int')
])([p1, p2])
lightgbm_init_kwargs = {
'boosting_type': Choice(['gbdt', 'dart', 'goss']),
'num_leaves': Choice([11, 31, 101, 301, 501]),
'learning_rate': Real(0.001, 0.1, step=0.005),
'n_estimators': 100,
'max_depth': -1,
'tree_learner': 'data' # add for dask
# subsample_for_bin = 200000, objective = None, class_weight = None,
# min_split_gain = 0., min_child_weight = 1e-3, min_child_samples = 20,
}
lightgbm_est = LightGBMDaskEstimator(task='binary',
fit_kwargs=lightgbm_fit_kwargs,
**lightgbm_init_kwargs)
xgb_init_kwargs = {
'tree_method': 'approx' # add for dask
}
xgb_est = XGBoostDaskEstimator(task='binary',
fit_kwargs=xgb_fit_kwargs,
**xgb_init_kwargs)
# catboost_init_kwargs = {
# 'silent': True
# }
# catboost_est = CatBoostEstimator(task='binary', fit_kwargs=catboost_fit_kwargs, **catboost_init_kwargs)
# or_est = ModuleChoice([lightgbm_est, xgb_est, catboost_est], name='estimator_options')(p3)
or_est = ModuleChoice([lightgbm_est, xgb_est],
name='estimator_options')(p3)
space.set_inputs(input)
return space
def get_space_column_transformer():
space = HyperSpace()
with space.as_default():
input = HyperInput(name='input1')
p1 = Pipeline(
[SimpleImputer(name='imputer1'),
StandardScaler(name='scaler1')],
columns=['a', 'b', 'c'],
name='p1')(input)
p2 = Pipeline(
[SimpleImputer(name='imputer2'),
StandardScaler(name='scaler2')],
columns=['c', 'd'],
name='p2')(input)
p3 = ColumnTransformer()([p1, p2])
space.set_inputs(input)
return space
def get_space_num_cat_pipeline_multi_complex(dataframe_mapper_default=False,
lightgbm_fit_kwargs={},
xgb_fit_kwargs={}):
space = HyperSpace()
with space.as_default():
input = HyperInput(name='input1')
p1 = numeric_pipeline_complex()(input)
p2 = categorical_pipeline_complex()(input)
p3 = DataFrameMapper(default=dataframe_mapper_default,
input_df=True,
df_out=True,
df_out_dtype_transforms=[(column_object,
'category')])([p1, p2])
p4 = numeric_pipeline_complex(seq_no=1)(p3)
p5 = categorical_pipeline_complex(seq_no=1)(p3)
p6 = DataFrameMapper(default=dataframe_mapper_default,
input_df=True,
df_out=True,
df_out_dtype_transforms=[(column_object,
'category')])([p4, p5])
lightgbm_init_kwargs = {
'boosting_type': Choice(['gbdt', 'dart', 'goss']),
'num_leaves': Choice([11, 31, 101, 301, 501]),
'learning_rate': Real(0.001, 0.1, step=0.005),
'n_estimators': 100,
'max_depth': -1,
# subsample_for_bin = 200000, objective = None, class_weight = None,
# min_split_gain = 0., min_child_weight = 1e-3, min_child_samples = 20,
}
lightgbm_est = LightGBMEstimator(task='binary',
fit_kwargs=lightgbm_fit_kwargs,
**lightgbm_init_kwargs)
xgb_init_kwargs = {}
xgb_est = XGBoostEstimator(task='binary',
fit_kwargs=xgb_fit_kwargs,
**xgb_init_kwargs)
or_est = ModuleChoice([lightgbm_est, xgb_est])(p6)
space.set_inputs(input)
return space
def __call__(self, *args, **kwargs):
space = HyperSpace()
with space.as_default():
hyper_input = HyperInput(name='input1')
estimators = []
if self.enable_dt:
estimators.append(self.dt)
if self.enable_dtr:
estimators.append(self.dtr)
if self.enable_lr:
estimators.append(self.lr)
if self.enable_nn:
estimators.append(self.nn)
modules = [ModuleSpace(name=f'{e["cls"].__name__}', **e) for e in estimators]
outputs = ModuleChoice(modules)(hyper_input)
space.set_inputs(hyper_input)
return space