def handle_transform(self, data_container: DataContainer,
context: ExecutionContext):
"""
Transform the data with the unions. It will make use of some parallel processing.
:param data_container: data container
:param context: execution context
:return: the transformed data_inputs.
"""
if self.n_jobs != 1:
data_containers = Parallel(
backend=self.backend,
n_jobs=self.n_jobs)(delayed(step.handle_transform)(
data_container.copy(), context.push(step))
for _, step in self.steps_as_tuple)
else:
data_containers = [
step.handle_transform(data_container.copy(),
context.push(step))
for _, step in self.steps_as_tuple
]
new_current_ids = self.hash(data_container)
data_container = self.joiner.handle_transform(data_containers,
new_current_ids)
return data_container
def _transform_data_container(self, data_container: DataContainer, context: ExecutionContext) -> DataContainer:
if any(not isinstance(di, DataContainer) for di in data_container.data_inputs):
raise ValueError("data_inputs given to ZipFeatures must be a list of DataContainer instances")
data_container = ZipDataContainer.create_from(*data_container.data_inputs)
if self.concatenate_inner_features:
data_container.concatenate_inner_features()
return data_container
def _transform_data_container(self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
"""
According to the idiom of `(1, 2, reversed(1))`, we do this, in order:
- `1`. Transform preprocessing step
- `2`. Transform postprocessing step
- `reversed(1)`. Inverse transform preprocessing step
:param data_container: data container to transform
:param context: execution context
:return: data_container
"""
data_container = self["preprocessing_step"].handle_transform(
data_container, context.push(self["preprocessing_step"]))
data_container = self["postprocessing_step"].handle_transform(
data_container, context.push(self["postprocessing_step"]))
data_container = self["preprocessing_step"].handle_inverse_transform(
data_container, context.push(self["preprocessing_step"]))
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
return data_container
def handle_fit(self, data_container: DataContainer,
context: ExecutionContext):
"""
Fit the parallel steps on the data. It will make use of some parallel processing.
:param data_container: The input data to fit onto
:param context: execution context
:return: self
"""
# Actually fit:
if self.n_jobs != 1:
fitted_steps_data_containers = Parallel(
backend=self.backend,
n_jobs=self.n_jobs)(delayed(step.handle_fit)(
data_container.copy(), context.push(step))
for _, step in self.steps_as_tuple)
else:
fitted_steps_data_containers = [
step.handle_fit(data_container.copy(), context.push(step))
for _, step in self.steps_as_tuple
]
# Save fitted steps
for i, (fitted_step, _) in enumerate(fitted_steps_data_containers):
self.steps_as_tuple[i] = (self.steps_as_tuple[i][0], fitted_step)
self._refresh_steps()
return self, data_container
def handle_fit_transform(self, data_container: DataContainer, context: ExecutionContext) -> (
'ReversiblePreprocessingWrapper', DataContainer):
"""
According to the idiom of `(1, 2, reversed(1))`, we do this, in order:
- `1`. Fit Transform preprocessing step
- `2`. Fit Transform postprocessing step
- `reversed(1)`. Inverse transform preprocessing step
:param data_container: data container to transform
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: (self, data_container)
:rtype: (ReversiblePreprocessingWrapper, DataContainer)
"""
self["preprocessing_step"], data_container = self["preprocessing_step"].handle_fit_transform(data_container,
context.push(self[
"preprocessing_step"]))
self["postprocessing_step"], data_container = self["postprocessing_step"].handle_fit_transform(data_container,
context.push(
self[
"postprocessing_step"]))
data_container = self["preprocessing_step"].handle_inverse_transform(data_container,
context.push(self["preprocessing_step"]))
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
return self, data_container
def _will_process(
self, data_container: DataContainer,
context: ExecutionContext) -> ('BaseTransformer', DataContainer):
"""
Flatten data container before any processing is done on the wrapped step.
:param data_container: data container to flatten
:param context: execution context
:return: (data container, execution context)
:rtype: ('BaseTransformer', DataContainer)
"""
data_container, context = super()._will_process(
data_container, context)
if data_container.expected_outputs is None:
expected_outputs = np.empty_like(
np.array(data_container.data_inputs))
expected_outputs.fill(np.nan)
data_container.set_expected_outputs(expected_outputs)
di, self.len_di = self._flatten_list(data_container.data_inputs)
eo, self.len_eo = self._flatten_list(data_container.expected_outputs)
flattened_data_container = DataContainer(
summary_id=data_container.summary_id,
data_inputs=di,
expected_outputs=eo,
sub_data_containers=data_container.sub_data_containers)
return flattened_data_container, context
def handle_transform(self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
data_container = FeatureUnion.handle_transform(self, data_container,
context)
results = self.judge.transform(data_container.data_inputs)
data_container.set_data_inputs(results)
return data_container
def _transform_data_container(self, data_container, context):
"""
Handle transform.
:param data_container: the data container to join
:param context: execution context
:return: transformed data container
"""
data_inputs = self.transform([dc.data_inputs for dc in data_container.data_inputs])
data_container = DataContainer(data_inputs=data_inputs, current_ids=data_container.current_ids,
expected_outputs=data_container.expected_outputs)
data_container.set_data_inputs(data_inputs)
return data_container
def handle_fit_transform(self, data_container: DataContainer,
context: ExecutionContext):
"""
Change the shape of the data container.
and/or
Apply any side effects based on the data container
And/or
Change the execution flow of the pipeline
"""
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
return self, data_container
def handle_transform(self, data_container: DataContainer,
context: ExecutionContext):
"""
Transform each step for each data inputs.
:param data_container: data container
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self
"""
output_data_container = ListDataContainer.empty()
for current_id, di, eo in data_container:
output = self.wrapped.handle_transform(
DataContainer(current_ids=None,
data_inputs=di,
expected_outputs=eo), context)
output_data_container.append(current_id, output.data_inputs,
output.expected_outputs)
current_ids = self.hash(data_container)
output_data_container.set_current_ids(current_ids)
return output_data_container
def _fit_transform_data_container(
self, data_container: DataContainer,
context: ExecutionContext) -> Tuple[BaseStep, DataContainer]:
"""
Fit transform each step for each data inputs, and expected outputs
:param data_container: data container to fit transform
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self, transformed_data_container
"""
output_data_container: DataContainer = ListDataContainer.empty(
original_data_container=data_container)
for current_id, di, eo in data_container:
try:
self.wrapped, output = self.wrapped.handle_fit_transform(
DataContainer(data_inputs=di,
current_ids=None,
expected_outputs=eo), context)
output_data_container.append(current_id, output.data_inputs,
output.expected_outputs)
except ContinueInterrupt:
continue
except BreakInterrupt:
break
output_data_container.summary_id = data_container.summary_id
return self, output_data_container
def _transform_data_container(self, data_container, context):
"""
Transform the data with the unions. It will make use of some parallel processing.
:param data_container: data container
:param context: execution context
:return: the transformed data_inputs.
"""
if self.n_jobs != 1:
data_containers = Parallel(
backend=self.backend,
n_jobs=self.n_jobs)(delayed(step.handle_transform)(
data_container.copy(), context)
for _, step in self.steps_as_tuple[:-1])
else:
data_containers = [
step.handle_transform(data_container.copy(), context)
for _, step in self.steps_as_tuple[:-1]
]
return DataContainer(
data_inputs=data_containers,
current_ids=data_container.current_ids,
summary_id=data_container.summary_id,
expected_outputs=data_container.expected_outputs,
sub_data_containers=data_container.sub_data_containers)
def _transform_data_container(self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
"""
Transform each step for each data inputs.
:param data_container: data container
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self
"""
output_data_container: ListDataContainer = ListDataContainer.empty(
original_data_container=data_container)
for current_id, di, eo in data_container:
output: DataContainer = self.wrapped.handle_transform(
DataContainer(data_inputs=di,
current_ids=None,
expected_outputs=eo), context)
output_data_container.append(current_id, output.data_inputs,
output.expected_outputs)
output_data_container.summary_id = data_container.summary_id
return output_data_container
def _fit_transform_data_container(self, data_container: DataContainer,
context: ExecutionContext):
"""
Fit transform each step for each data inputs, and expected outputs
:param data_container: data container to fit transform
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self, transformed_data_container
"""
output_data_container = ListDataContainer.empty()
for current_id, di, eo in data_container:
self.wrapped, output = self.wrapped.handle_fit_transform(
DataContainer(current_ids=None,
data_inputs=di,
expected_outputs=eo), context)
output_data_container.append(current_id, output.data_inputs,
output.expected_outputs)
output_data_container.summary_id = data_container.summary_id
return self, output_data_container
def _create_expanded_data_container(
self, data_container: DataContainer) -> ExpandedDataContainer:
"""
Create expanded data container.
:param data_container: data container to expand
:type data_container: DataContainer
:return: expanded data container
:rtype: ExpandedDataContainer
"""
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
expanded_data_container = ExpandedDataContainer.create_from(
data_container)
return expanded_data_container
def _transform_data_container(self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
"""
Nullify wrapped step hyperparams, and don't transform the wrapped step.
:param data_container: data container
:param context: execution context
:return: step, data_container
"""
if self.hyperparams[OPTIONAL_ENABLED_HYPERPARAM]:
return self.wrapped.handle_transform(data_container, context)
self._nullify_hyperparams()
data_container.set_data_inputs(self.nullified_return_value)
return DataContainer(data_inputs=self.nullified_return_value,
current_ids=data_container.current_ids,
expected_outputs=self.nullified_return_value)
def handle_fit(self, data_container: DataContainer, context: ExecutionContext) -> 'ReversiblePreprocessingWrapper':
"""
Handle fit by fitting preprocessing step, and postprocessing step.
:param data_container: data container to fit on
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self, data_container
:rtype: (ReversiblePreprocessingWrapper, DataContainer)
"""
self["preprocessing_step"], data_container = \
self["preprocessing_step"].handle_fit_transform(data_container, context.push(self["preprocessing_step"]))
self["postprocessing_step"] = \
self["postprocessing_step"].handle_fit(data_container, context.push(self["postprocessing_step"]))
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
return self
def _did_process(self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
"""
Reaugment the flattened data container.
:param data_container: data container to then_unflatten
:param context: execution context
:return: data container
"""
data_container = super()._did_process(data_container, context)
if self.then_unflatten:
data_container.set_data_inputs(
self._reaugment_list(data_container.data_inputs, self.len_di))
data_container.set_expected_outputs(
self._reaugment_list(data_container.expected_outputs,
self.len_eo))
self.len_di = []
self.len_eo = []
return data_container
def _fit_data_container(
self, data_container: DataContainer,
context: ExecutionContext) -> ('BaseStep', DataContainer):
fitted_steps = []
for i, (current_ids, data_inputs,
expected_outputs) in enumerate(data_container):
fitted_step = self.steps[i].handle_fit(
DataContainer(current_ids=current_ids,
data_inputs=data_inputs,
expected_outputs=expected_outputs), context)
fitted_steps.append(fitted_step)
self.steps = fitted_steps
return self
def _transform_data_container(self, data_container: DataContainer,
context: ExecutionContext):
data_containers = list(
filter(
lambda dc:
(len(dc.data_inputs) > 0 and len(dc.expected_outputs) > 0),
data_container.data_inputs))
if len(data_containers) == 1:
return data_containers[0]
else:
return DataContainer(
data_inputs=list(map(attrgetter("data_inputs"))),
expected_outputs=list(map(attrgetter("expected_outputs"))),
current_ids=data_container.current_ids)
def _inverse_transform_data_container(
self, data_container: DataContainer,
context: ExecutionContext) -> DataContainer:
inverse_transform_results = []
for i, (current_ids, data_inputs,
expected_outputs) in enumerate(data_container):
inverse_transform_result = self[i].handle_inverse_transform(
DataContainer(current_ids=current_ids,
data_inputs=data_inputs,
expected_outputs=expected_outputs), context)
inverse_transform_results.append(inverse_transform_result)
output_data_container = ListDataContainer.empty()
for data_container_batch in inverse_transform_results:
output_data_container.append_data_container(data_container_batch)
return output_data_container
def _fit_data_container(self, data_container: DataContainer,
context: ExecutionContext) -> BaseStep:
"""
Fit each step for each data inputs, and expected outputs
:param data_container: data container
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: self
"""
for current_id, di, eo in data_container:
self.wrapped = self.wrapped.handle_fit(
DataContainer(data_inputs=di,
current_ids=None,
expected_outputs=eo), context)
return self
def _fit_transform_data_container(
self, data_container: DataContainer,
context: ExecutionContext) -> ('BaseStep', DataContainer):
fitted_steps_data_containers = []
for i, (current_ids, data_inputs,
expected_outputs) in enumerate(data_container):
fitted_step_data_container = self.steps[i].handle_fit_transform(
DataContainer(current_ids=current_ids,
data_inputs=data_inputs,
expected_outputs=expected_outputs), context)
fitted_steps_data_containers.append(fitted_step_data_container)
self.steps = [step for step, _ in fitted_steps_data_containers]
output_data_container = ListDataContainer.empty()
for _, data_container_batch in fitted_steps_data_containers:
output_data_container.append_data_container(data_container_batch)
return self, output_data_container
def handle_fit_transform(self, data_container: DataContainer, context: ExecutionContext) -> (
'BaseStep', DataContainer):
"""
Nullify wrapped step hyperparams, and don't fit_transform the wrapped step.
:param data_container: data container
:type data_container: DataContainer
:param context: execution context
:type context: ExecutionContext
:return: step, data_container
:type: (BaseStep, DataContainer)
"""
if self.hyperparams[OPTIONAL_ENABLED_HYPERPARAM]:
self.wrapped, data_container = self.wrapped.handle_fit_transform(data_container, context)
return self, data_container
self._nullify_hyperparams()
return self, DataContainer(data_inputs=self.nullified_return_value, current_ids=data_container.current_ids,
expected_outputs=self.nullified_return_value)
def _transform_data_container(self, data_container: DataContainer,
context: ExecutionContext):
"""
Handle transform.
:param data_container: the data container to join
:param context: execution context
:return: transformed data container
"""
data_inputs = [
dc.data_inputs for dc in data_container.data_inputs
if len(dc.data_inputs) > 0
]
if len(data_inputs) == 1:
data_inputs = data_inputs[0]
data_container = DataContainer(
data_inputs=data_inputs,
current_ids=data_container.current_ids,
expected_outputs=data_container.expected_outputs)
return data_container
def handle_transform(self, data_containers,
context: ExecutionContext) -> DataContainer:
"""
Handle transform.
:param data_containers: the data container to join
:param context: execution context
:return: transformed data container
"""
data_inputs = self.transform(
[dc.data_inputs for dc in data_containers])
data_container = DataContainer(
current_ids=data_containers[-1].current_ids,
data_inputs=data_inputs,
expected_outputs=data_containers[-1].expected_outputs)
data_container.set_data_inputs(data_inputs)
current_ids = self.hash(data_container)
data_container.set_current_ids(current_ids)
return data_container
def _will_process(
self, data_container: DataContainer,
context: ExecutionContext) -> (DataContainer, ExecutionContext):
return data_container.to_numpy(), context
def _did_process(self, data_container: DataContainer,
context: ExecutionContext):
data_container = super()._did_process(data_container, context)
return data_container.reduce_dim()
