ballet.pipeline module¶

class ballet.pipeline.EngineerFeaturesResult(X_df, features, pipeline, X, y_df, encoder, y)[source]¶

Bases: tuple

X: numpy.ndarray¶: Alias for field number 3

X_df: pandas.core.frame.DataFrame¶: Alias for field number 0

encoder: ballet.eng.base.BaseTransformer¶: Alias for field number 5

features: List[ballet.feature.Feature]¶: Alias for field number 1

pipeline: ballet.pipeline.FeatureEngineeringPipeline¶: Alias for field number 2

y: numpy.ndarray¶: Alias for field number 6

y_df: pandas.core.frame.DataFrame¶: Alias for field number 4

class ballet.pipeline.FeatureEngineeringPipeline(features)[source]¶

Bases: sklearn_pandas.dataframe_mapper.DataFrameMapper

Feature engineering pipeline

Parameters: features (Union[Feature, List[Feature]]) – feature or list of features

property ballet_features¶

Return type: List[Feature]

get_names(columns, transformer, x, alias=None)[source]¶

Return verbose names for the transformed columns.

This extends the behavior of DataFrameMapper to allow alias to rename all of the output columns, rather than just providing a common base. It also allows columns to be a callable that supports selection by callable of the data frame.

ballet.pipeline.make_engineer_features(pipeline, encoder, load_data)[source]¶

Return type: Callable[[DataFrame, DataFrame], EngineerFeaturesResult]