seldonian.dataset.SupervisedDataSet

class SupervisedDataSet(features, labels, sensitive_attrs, num_datapoints, meta)

Bases: DataSet

__init__(features, labels, sensitive_attrs, num_datapoints, meta)

Class for holding supervised learning data and metadata.

Parameters:

  • features (numpy.ndarray or list) – Feature array - 2D array of shape (num_datapoints,n_features) or a list of 1D feature columns

  • labels (numpy.ndarray) – Labels for each data point. Only support 1D arrays.

  • sensitive_attrs (numpy.ndarray, defaults to []) – Sensitive attribute array for each data point

  • num_datapoints (int) – Number of data points in the dataset

  • meta (MetaData) – Metadata object

__repr__()

Return repr(self).

Methods