syne_tune.optimizer.schedulers.searchers.conformal.surrogate.surrogate_model module

class syne_tune.optimizer.schedulers.searchers.conformal.surrogate.surrogate_model.SurrogateModel(config_space, mode, max_fit_samples=None, random_state=None)[source]

Bases: object

suggest(replace_config=False)[source]

Return type:: dict

fit(df_features, y, ncandidates=2000)[source]

predict(df_features)[source]

Parameters:: df_features (DataFrame) – input features to make predictions with shape (n, d)
Return type:: Tuple[array, array]
Returns:: predictions in the shape of (n,)

# TODO should we rather have (n, 1)? input is taken in this form