Experiment

class Experiment(group_name: str, workflow: str, algorithms: Dict[str, AlgorithmWrapper], dataset_path: Path, workflow_args: Dict[str, Any], split_index: int, table_name: str | None, categorical_features: List[str] | None)

Configuration for a single experiment run.

Parameters:

group_namestr

Name of the experiment group for organization

algorithmsdict

Dictionary of AlgorithmWrapper instances with standardized keys:

• Single modeldict

  {“model”: wrapper}

• Multiple modelsdict

  {“model”: wrapper1, “model2”: wrapper2, …}

dataset_pathPath or str

Path to the dataset file

workflow_argsdict

Arguments to pass to the workflow

split_index: int

Index value used to select a DataSplitInfo instance from DataSplits

table_namestr, optional

Name of table for database files

categorical_featureslist of str, optional

Names of categorical features in the dataset

Attributes:

namestr

Generate full descriptive name for logging and debugging.

workflow: str

Name of the workflow file to use (without .py extension)

dataset_nameTuple[str, Optional[str]]

Get the dataset name with optional table name.

table_name: str

Name of the table (database files only)

algorithm_kwargsdict

Get dictionary of instantiated algorithms.

algorithm_nameslist

Get list of algorithm names.

workflow_attributesdict

Get combined workflow and algorithm arguments to pass to Workflow.

property algorithm_kwargs: dict

Get dictionary of instantiated algorithms.

Returns:

dict

Mapping of keys to algorithm instances Example: {‘model’: LinearRegression()}

property algorithm_names: list

Get list of algorithm names.

Returns:

list

Names of all algorithms in this experiment Example: [‘linear’, ‘ridge’]

property dataset_name: Tuple[str, str | None]

Get the dataset name with optional table name.

Returns:

str

Dataset stem with optional table name Example: ‘data_table1’ or ‘data’

property name: str

Generate full descriptive name for logging and debugging.

Returns:

str

Name combining group name and algorithm names Example: ‘baseline_linear_ridge’

property workflow_attributes: Dict[str, Any]

Get combined workflow and algorithm arguments to pass to Workflow.

Returns:

dict

Union of workflow_args and algorithm_kwargs