Model: Base

class xaicompare.adapters.models.model_base.ModelAdapter(model, class_names: Sequence[str] | None = None)[source]

Bases: object

build_text_index(X_test, y_test: Sequence | None = None, raw_text: Sequence | None = None, class_names: Sequence[str] | None = None, **kwargs: Any) → DataFrame[source]
Return a DataFrame with at least:

  • id (int)

  • text (str) # original doc text if available

  • y_true (optional)

  • y_pred (optional)

  • proba_{class} (optional per class)

Default impl: try best-effort; subclasses can override.

class_names() → List[str][source]

Human-readable class names.

feature_names() → List[str][source]

Vectorizer / model feature names (tokens, n-grams).

is_sparse_input() → bool[source]
predict(X: ndarray | spmatrix | list) → ndarray[source]
predict_proba(X: ndarray | spmatrix | list) → ndarray | None[source]

Return (n_samples, n_classes) or None if not supported.