Estimator types#
This section lists the various object types (scitypes) available in sktime.
sktime supports unified interfaces for different types of algorithms and objects.
These are internally referred to as scitypes, with strict string identifiers
such as "forecaster", "classifier", "detector", etc.
Each object type corresponds to a specific unified interface and base class.
Every object in sktime has one or more scitypes, which can be inspected
via its "object_type" tag.
For a list of all tags, see Object and estimator tags.
AI algorithms#
Time series aligner or sequence aligner. |
|
Time series classifier. |
|
Time series clusterer. |
|
Detector of anomalies, outliers, or change points. |
|
Time series forecaster. |
|
Parameter fitting estimator. |
|
Time series regressor. |
|
Time series transformer. |
|
Pairwise transformer for tabular data, distance or kernel. |
|
Pairwise transformer for panel data, distance or kernel. |
Data handling#
Dataset object. |
|
Classification Dataset. |
|
Forecasting Dataset class. |
|
Regression Dataset class. |
|
Time series splitter. |
Evaluation and metrics#
Performance metric for time series. |
|
Performance metric for time series detection tasks. |
|
Performance metric for time series forecasting, point forecasts. |
|
Performance metric for time series forecasting, probabilistic forecasts. |
Catalogues and collections#
Catalogue of objects. |
General types#
Retrieving estimator types programmatically#
|
Return list of object scitype shorthands in sktime. |
|
Return register of object scitypes and base classes in sktime. |
|
Return lookup dictionary of scitype shorthands to base classes in sktime. |