def get_aggregate_metrics()

in src/entrypoint/gluonts_example/evaluator.py [0:0]

• 11 lines of code
• 3 McCabe index (conditional complexity)

    def get_aggregate_metrics(self, metric_per_ts: pd.DataFrame) -> Tuple[Dict[str, float], pd.DataFrame]:
        totals, metrics_per_ts = super().get_aggregate_metrics(metric_per_ts)

        # region: for each metric, aggregate across timeseries.
        # Aggregation step
        agg_funs = {
            "wMAPE": "mean",
        }
        assert set(metric_per_ts.columns) >= agg_funs.keys(), "The some of the requested item metrics are missing."
        my_totals = {key: metric_per_ts[key].agg(agg) for key, agg in agg_funs.items()}

        # Update base metrics with our custom metrics.
        totals.update(my_totals)
        # endregion

        # Save montage
        self.mp.savefig()

        # Make sure to flush buffered results to the disk.
        self.out_f.close()

        return totals, metrics_per_ts