BlueprintConverter#

class datarobotx.BlueprintConverter#

Inherited attributes:

`CATEGORICAL_COLUMN_TEMPLATE`
`DEFAULT_IMPORTS`
`STANDARD_INPUT_TEMPLATE`

Methods:

`build_blueprint_imports`(stages)	rtype: `List`[`str`]
`build_preprocessor_pipeline`(stages, ...)	Build a preprocessor pipeline from blueprint stages.
`construct_stage_input_paths`(stages, ...)	rtype: `List`[`List`[`str`]]
`convert`(blueprint_json)	Convert a blueprint to an open source representation block of code.
`convert_blueprint_class_to_open_representation`(...)	rtype: `Tuple`[`str`, `str`, `Union`[`str`, `List`[`str`], `None`]]
`extract_dr_class_params_from_blueprint_parts`(...)	Extract the input parameters and values from the blueprint class representation.
`extract_element_from_stage`(stage)	rtype: `Tuple`[`str`, `str`]
`extract_estimator_inputs`(blueprint_json)	rtype: `Tuple`[`List`[`str`], `str`]
`extract_stages`(blueprint_json)	rtype: `Dict`[`str`, `StageDict`]
`parse_blueprint_stage`(blueprint_stage)	Convert a stage of a blueprint to an open representation mapping.

Inherited methods:

`__init__`()
`format_code_block_with_black`(code_block)	rtype: `str`
`make_blueprint`(blueprint_imports, ...)	Construct and format code to create a code representation of a blueprint.
`make_column_transformer_element_from_pipeline`(...)	Construct a column tranformer element from a pre-formatted transformer pipeline.
`make_column_transformer_from_column_transformer_elements`(...)	Return a Column Transformer representation from a pre-formatted transformer pipeline The transformer should be ultimately constructed from: make_column_transformer_element_from_pipeline_elements If starting from steps, see make_pipeline_elements_from_steps.
`make_pipeline_elements_from_steps`(pipeline_steps)	Build a string of pipeline elements from raw pipeline steps.
`make_pipeline_from_pipeline_elements`(...)	Return a Pipeline representation from a pre-formatted transformer pipeline element Elements should be created from steps with make_pipeline_elements_from_steps.
`make_standard_template`()	Construct a standard template for the top of the code.

classmethod build_preprocessor_pipeline(stages, stage_input_paths)#

Build a preprocessor pipeline from blueprint stages.

Return type:: List[str]

classmethod convert(blueprint_json)#

Convert a blueprint to an open source representation block of code.

Parameters:: blueprint_json (Union[BlueprintJson, str]) – A json formatted dictionary of a datarobot blueprint. Can be retrieved using the datarobot API. This can also be passed as the string that represents the json formatted dictionary directly.
Return type:: str

static extract_dr_class_params_from_blueprint_parts(blueprint_parts)#

Extract the input parameters and values from the blueprint class representation.

Return type:: Dict[str, str]

classmethod make_blueprint(blueprint_imports, preprocessor_pipeline, estimator_element)#

Construct and format code to create a code representation of a blueprint.

Parameters:

blueprint_imports (list[str]) – List of imports that should be added to the header
preprocessor_pipeline (list[str]) – Column Transformer code string. See make_column_transformer_from_pipeline
estimator_element (Tuple[str, str]) – A tuples represented as (estimator_name, estimator_action). The step_action can either be a class call (e.g. sklearn.impute.SimpleImputer()) or “passthrough”. The step_action is typically generated by the MappingObject.get_open_representation() function.

Return type:

str

classmethod make_column_transformer_element_from_pipeline(transformer_name, transformer_pipeline, transformer_columns=None)#

Construct a column tranformer element from a pre-formatted transformer pipeline.

Pipeline should be created from steps with: make_pipeline_elements_from_steps and make_pipeline_from_pipeline_elements

Return type:: str

static make_column_transformer_from_column_transformer_elements(column_transformer_elements)#

Return a Column Transformer representation from a pre-formatted transformer pipeline The transformer should be ultimately constructed from: make_column_transformer_element_from_pipeline_elements If starting from steps, see make_pipeline_elements_from_steps.

Return type:: str

static make_pipeline_elements_from_steps(pipeline_steps, reverse=False, index=None)#

Build a string of pipeline elements from raw pipeline steps.

Parameters:

pipeline_steps (List[Tuple[str, str]]) – A list of tuples, where each tuple is (step_name, step_action). The step_action can either be a class call (e.g. sklearn.impute.SimpleImputer()) or “passthrough”. The step_action is typically generated by the MappingObject.get_open_representation() function.
reverse (bool) – Reverse the list. If steps are added from last to first, this should be set to True so that the final pipeline is in the correct direction.

Return type:

str

static make_pipeline_from_pipeline_elements(pipeline_element_str)#

Return a Pipeline representation from a pre-formatted transformer pipeline element Elements should be created from steps with make_pipeline_elements_from_steps.

Return type:: str

classmethod make_standard_template()#

Construct a standard template for the top of the code.

Return type:: str

classmethod parse_blueprint_stage(blueprint_stage)#

Convert a stage of a blueprint to an open representation mapping.

Return type:: StageDict