gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.performance module#

class gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.performance.Performance(training_time: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, latency: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, num_model_parameters: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, num_gradient_updates: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, ncrps: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, mase: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, smape: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, nrmse: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric, nd: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric)[source]#

Bases: object

The performance class encapsulates the metrics that are recorded for configurations.

classmethod from_dict(metrics: dict[str, float | int]) → Performance[source]#

Initializes a new performance object from the given 1D dictionary.

Metrics are expected to be provided via <metric>_mean and <metric>_std keys.

latency: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

mase: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

classmethod metrics() → list[str][source]#: Returns the list of metrics that are exposed by the performance class.

ncrps: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

nd: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

nrmse: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

num_gradient_updates: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

num_model_parameters: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

smape: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#

classmethod to_dataframe(performances: list[Performance], std: bool = True) → pd.DataFrame[source]#: Returns a data frame representing the provided performances.

training_time: gluonts.nursery.tsbench.src.tsbench.evaluations.metrics.metric.Metric#