xai_compare.factories.factory

Classes

ComparisonFactory(model, data, target[, ...])	A class for creating classes or instances of various comparison techniques based on the given explainers.
ExplainerFactory([model, X_train, X_test, ...])	A class for creating instances of various explainer types based on the given model and data.

class xai_compare.factories.factory.ComparisonFactory(model, data: DataFrame, target: DataFrame | Series | ndarray, mode: str = 'regression', random_state: int = 42, verbose: bool = True, threshold: float = 0.2, metric: str | None = None, n_splits: int = 5, default_explainers: List[str] = ['shap', 'lime', 'permutations'], custom_explainer: Type[Explainer] | List[Type[Explainer]] | None = None)

A class for creating classes or instances of various comparison techniques based on the given explainers.

This class simplifies the process of instantiating different types of comparison techniques by providing a common interface to specify the necessary training and testing datasets along with the model.

create(comparison_type: <module 'string' from '/home/docs/.asdf/installs/python/3.10.14/lib/python3.10/string.py'>)

Creates and returns a comparison object based on the specified type.

This method allows for dynamic creation of different types of comparison objects based on a predefined string identifier. If the model is not set, it returns the class of the comparison for manual instantiation.

Attributes:

comparison_type (str):

A string identifier for the comparison type. Valid options are “feature_selection”, “consistency”.

Returns:

comparison.Comparison:

An instance or class of the requested comparison type with the provided model and data.

Raises:

ValueError:

If the comparison_type is not recognized or unsupported.

class xai_compare.factories.factory.ExplainerFactory(model: Any | None = None, X_train: DataFrame | None = None, X_test: DataFrame | None = None, y_train: DataFrame | Series | ndarray | None = None, y_test: DataFrame | Series | ndarray | None = None, mode: str = 'regression')

A class for creating instances of various explainer types based on the given model and data.

This class simplifies the process of instantiating different types of explainers by providing a common interface to specify the necessary training and testing datasets along with the model.

Attributes:

model (Any):

The machine learning model to be explained.

X_train (pd.DataFrame):

Training features dataset.

X_test (pd.DataFrame):

Testing features dataset.

y_train (Union[pd.DataFrame, pd.Series, np.ndarray]):

Training target dataset.

y_test (Union[pd.DataFrame, pd.Series, np.ndarray]):

Testing target dataset.

mode (str):

Mode of operation, specifies if the model is for ‘regression’ or ‘classification’.

create(explainer_type: <module 'string' from '/home/docs/.asdf/installs/python/3.10.14/lib/python3.10/string.py'>) → Explainer

Creates and returns an explainer object based on the specified type.

This method allows for dynamic creation of different types of explainer objects based on a predefined string identifier. If the model is not set, it returns the class of the explainer for manual instantiation.

Attributes:

explainer_type (str):

A string identifier for the explainer type. Valid options are “shap”, “lime”, “permutations”.

Returns:

Explainer:

An instance of the requested explainer type with the provided model and data.

Raises:

ValueError:

If the explainer_type is not recognized or unsupported.