def _get_label_encoder_and_max()

in causalml/features.py [0:0]

• 12 lines of code
• 5 McCabe index (conditional complexity)

    def _get_label_encoder_and_max(self, x):
        """Return a mapping from values and its maximum of a column to integer labels.

        Args:
            x (pandas.Series): a categorical column to encode.

        Returns:
            label_encoder (dict): mapping from values of features to integers
            max_label (int): maximum label
        """

        # NaN cannot be used as a key for dict. So replace it with a random integer.
        label_count = x.fillna(NAN_INT).value_counts()
        n_uniq = label_count.shape[0]

        label_count = label_count[label_count >= self.min_obs]
        n_uniq_new = label_count.shape[0]

        # If every label appears more than min_obs, new label starts from 0.
        # Otherwise, new label starts from 1 and 0 is used for all old labels
        # that appear less than min_obs.
        offset = 0 if n_uniq == n_uniq_new else 1

        label_encoder = pd.Series(
            np.arange(n_uniq_new) + offset, index=label_count.index
        )
        max_label = label_encoder.max()
        label_encoder = label_encoder.to_dict()

        return label_encoder, max_label