bluecast.blueprints.orchestration

Module containing model orchestration tools.

Module Contents

Classes

ModelMatchMaker

Matching the incoming data with the best model based on the adversarial validation score.
class bluecast.blueprints.orchestration.ModelMatchMaker

Matching the incoming data with the best model based on the adversarial validation score.

append_model_and_dataset(bluecast_instance: bluecast.blueprints.cast.BlueCast | bluecast.blueprints.cast_regression.BlueCastRegression | bluecast.blueprints.cast_cv.BlueCastCV | bluecast.blueprints.cast_cv_regression.BlueCastCVRegression, df: pandas.DataFrame)

Append the model and the dataset to the matchmaker.

Parameters:

  • bluecast_instance – The BlueCast instance to append.

  • df – The dataset to append.

find_best_match(df: pandas.DataFrame, use_cols: List[int | float | str], cat_columns: List | None, delta: float, train_on_device: str = 'cpu') → Tuple[bluecast.blueprints.cast.BlueCast | bluecast.blueprints.cast_regression.BlueCastRegression | bluecast.blueprints.cast_cv.BlueCastCV | bluecast.blueprints.cast_cv_regression.BlueCastCVRegression | None, pandas.DataFrame | None]

Find the best match based on the adversarial validation score. :param df: Dataset to match. :param use_cols: Columns to use for the adversarial validation. Numerical columns are allowed only. :param delta: Maximum delta for the adversarial validation score to be away from 0.5. If no dataset reaches this

delta, (None, None) is returned.

Parameters:

  • cat_columns – (Optional) List with names of categorical columns.

  • train_on_device – Device to train the model on. Options are ‘cpu’ and ‘gpu’. (Default is ‘cpu’)

Returns:

If a match is found, the BlueCast instance and the dataset are returned. Otherwise, (None, None) is returned.