GraphNode

class fedot.core.pipelines.node.NodeMetadata(metric: float = None)

Bases: object

Parameters

metric (float) –

Return type

None

metric: float = None
class fedot.core.pipelines.node.Node(nodes_from, operation_type=None, log=None, **kwargs)

Bases: fedot.core.dag.graph_node.GraphNode

Base class for Node definition in Pipeline structure

Parameters

  • nodes_from (Optional[List[Node]]) – parent nodes which information comes from

  • operation_type (Union[str, Operation, None]) – str type of the operation defined in operation repository the custom prefix can be added after / (to highlight the specific node) The prefix will be ignored at Implementation stage

  • log (Optional[fedot.core.log.Log]) – Log object to record messages

update_params()
property operation
property fitted_operation
unfit()
fit(input_data)

Run training process in the node

Parameters

input_data (fedot.core.data.data.InputData) – data used for operation training

Return type

fedot.core.data.data.OutputData

predict(input_data, output_mode='default')

Run prediction process in the node

Parameters

  • input_data (fedot.core.data.data.InputData) – data used for prediction

  • output_mode (str) – desired output for operations (e.g. labels, probs, full_probs)

Return type

fedot.core.data.data.OutputData

property custom_params
property tags

Return tags of operation in the node.

class fedot.core.pipelines.node.PrimaryNode(operation_type=None, node_data=None, **kwargs)

Bases: fedot.core.pipelines.node.Node

The class defines the interface of Primary nodes where initial task data is located

Parameters

  • operation_type (Union[str, Operation, None]) – str type of the operation defined in operation repository

  • node_data (dict) – dictionary with InputData for fit and predict stage

  • kwargs – optional arguments (i.e. logger)

fit(input_data, **kwargs)

Fit the operation located in the primary node

Parameters

input_data (fedot.core.data.data.InputData) – data used for operation training

Return type

fedot.core.data.data.OutputData

unfit()
predict(input_data, output_mode='default')

Predict using the operation located in the primary node

Parameters

  • input_data (fedot.core.data.data.InputData) – data used for prediction

  • output_mode (str) – desired output for operations (e.g. labels, probs, full_probs)

Return type

fedot.core.data.data.OutputData

get_data_from_node()

Method returns data if the data was set to the nodes directly

property node_data
class fedot.core.pipelines.node.SecondaryNode(operation_type=None, nodes_from=None, **kwargs)

Bases: fedot.core.pipelines.node.Node

The class defines the interface of Secondary nodes modifying tha data flow in Pipeline

Parameters

  • operation_type (Union[str, Operation, None]) – str type of the operation defined in operation repository

  • nodes_from (Optional[List[Node]]) – parent nodes where data comes from

  • kwargs – optional arguments (i.e. logger)

fit(input_data, **kwargs)

Fit the operation located in the secondary node

Parameters

input_data (fedot.core.data.data.InputData) – data used for operation training

Return type

fedot.core.data.data.OutputData

predict(input_data, output_mode='default')

Predict using the operation located in the secondary node

Parameters

  • input_data (fedot.core.data.data.InputData) – data used for prediction

  • output_mode (str) – desired output for operations (e.g. labels, probs, full_probs)

Return type

fedot.core.data.data.OutputData

fedot.core.pipelines.node.get_default_params(model_name)
Parameters

model_name (str) –