Catalogues#

catalogues provide a unified interface to list, query, and retrieve collections of sktime objects such as datasets, estimators, metrics, and cross-validation strategies.

Catalogues can make reproducing and extending benchmarking studies and competitions easier and convenient, by reducing boilerplate code.

A catalogue can be used to:

  • inspect what items exist in a given collection,

  • retrieve items by category (e.g., only datasets, only metrics),

  • obtain either string specifications or fully constructed objects,

  • describe metadata about the catalogue such as number of items, authorship, and source.

Forecasting Catalogues#

dummy._dummy_forecasting.DummyForecastingCatalogue()

Dummy catalogue of datasets, forecasters, metrics, and cv.

Classification Catalogues#

dummy._dummy_classification.DummyClassificationCatalogue()

Dummy catalogue of datasets, classifiers, metrics, and cv.