fcb01e3b18c0b3304c4ff0b1f0b8b4ede518f762,ludwig/features/numerical_feature.py,NumericalFeatureMixin,add_feature_data,#Any#Any#Any#Any#Any#Any#,82

Before Change

        proc_df[feature[PROC_COLUMN]] = input_df[feature[COLUMN]].values
        if preprocessing_parameters["normalization"] is not None:
            if preprocessing_parameters["normalization"] == "zscore":
                mean = metadata[feature[NAME]]["mean"]
                std = metadata[feature[NAME]]["std"]
                proc_df[feature[PROC_COLUMN]] = \
                    (proc_df[feature[PROC_COLUMN]] - mean) / std
            elif preprocessing_parameters["normalization"] == "minmax":

After Change

            np.float32).values

        // normalize data as required
        numeric_transformer = get_from_registry(
            preprocessing_parameters.get("normalization", None),
            numeric_transformation_registry
        )(**metadata[feature[NAME]])

        proc_df[feature[PROC_COLUMN]] = \
            numeric_transformer.transform(proc_df[feature[PROC_COLUMN]])

Instances