Skip to content

AutoTabPFNClassifier

Bases: BaseEstimator, ClassifierMixin

Automatic Post Hoc Ensemble Classifier for TabPFN models.

Attributes:

Name Type Description
predictor_

AutoPostHocEnsemblePredictor The predictor interface used to make predictions, see post_hoc_ensembles.pfn_phe.AutoPostHocEnsemblePredictor for more.

phe_init_args_

dict The optional initialization arguments used for the post hoc ensemble predictor.

__init__

__init__(
    max_time: int | None = 30,
    preset: Literal[
        "default", "custom_hps", "avoid_overfitting"
    ] = "default",
    ges_scoring_string: str = "roc",
    device: Literal["cpu", "cuda"] = "cpu",
    random_state: int | None | RandomState = None,
    categorical_feature_indices: list[int] | None = None,
    phe_init_args: dict | None = None,
)

Parameters:

Name Type Description Default
max_time

int | None, default=None The maximum time to spend on fitting the post hoc ensemble.

30
preset Literal['default', 'custom_hps', 'avoid_overfitting']

{"default", "custom_hps", "avoid_overfitting"}, default="default" The preset to use for the post hoc ensemble.

'default'
ges_scoring_string

str, default="roc" The scoring string to use for the greedy ensemble search. Allowed values are: {"accuracy", "roc" / "auroc", "f1", "log_loss"}.

'roc'
device

{"cpu", "cuda"}, default="cuda" The device to use for training and prediction.

'cpu'
random_state

int, RandomState instance or None, default=None Controls both the randomness base models and the post hoc ensembling method.

None
categorical_feature_indices list[int] | None

list[int] or None, default=None The indices of the categorical features in the input data. Can also be passed to fit().

None
phe_init_args

dict | None, default=None The initialization arguments for the post hoc ensemble predictor. See post_hoc_ensembles.pfn_phe.AutoPostHocEnsemblePredictor for more options and all details.

None

fit

fit(
    X,
    y,
    categorical_feature_indices: list[int] | None = None,
)

predict

predict(X)

predict_proba

predict_proba(X)