Explainer Base

class xailens.adapters.explainers.xaimethod_base.XAIMethodAdapter

Bases: ABC

Abstract class for XAI method adapter.

_abc_impl = <_abc._abc_data object>

compatible_visualisations: dict | None

default_visualisation: str | None

display_name: str | None = None

display_notes: str | None = None

abstractmethod explain_global_df(ctx: ModelContext, model=None) → DataFrame

Computes the global explanation for the model

Parameters:

• ctx – ModelContext object

• model – Trained model object

Returns:

dataframe containing global explanation

Return type:

pd.DataFrame

Raises:

MissingDataError – if required data is not provided

abstractmethod explain_local_df(ctx: ModelContext, model=None) → DataFrame

Computes the local explanation for the model

Parameters:

• ctx – ModelContext object

• model – Trained model object

Returns:

dataframe containing local explanations

Return type:

pd.DataFrame

Raises:

MissingDataError – if required data is not provided

abstractmethod explain_local_json(ctx: ModelContext) → list[dict[str, Any]]

Computes the local explanation for the model

Parameters:

ctx – ModelContext object

Returns:

list of dicts - JSON containing local explanations

Raises:

MissingDataError – if required data is not provided

explanation_file_type: str = 'df'

get_compatible_visualisations()

Gets the compatible visualisations for the instantiated class

get_default_visualisation()

Gets the default visualisation for the instantiated class

classmethod load_global_explanation(model: Model) → DataFrame

Loads the global explanation for the instantiated class

Parameters:

model – Model object

Returns:

dataframe containing global explanation

Return type:

pd.DataFrame

classmethod load_local_explanation(model: Model, instance_id: str) → DataFrame