Object and estimator tags#

Every first-class object in skpro is tagged with a set of tags that describe its properties and capabilities, or control its behavior.

Tags are key-value pairs, where the key is a string with the name of the tag. The value of the tag can have arbitrary type, and describes a property, capability, or controls behaviour of the object, depending on the tag.

For instance, a probabilistic regressor has the tag "capability:survival": True if it can use censoring information for survival analysis. Users can find all regressors with this capability by filtering for this tag.

This API reference lists all tags available in skpro, and key utilities for their usage.

Inspecting tags, retrieving by tags#

Tags can be inspected at runtime using the following utilities:

to get the tags of an object, use the get_tags method. An object’s tags can depend on its hyper-parameters.
to get the tags of a class, use the get_class_tags method of the class. A class’s tags are static and do not depend on its hyper-parameters. By default, class tags that may vary for instances take the most “capable” value, in the case of capabilities.
to programmatically retrieve all tags available in skpro or for a particular type of object, at runtime, use the registry.all_tags utility.

`all_tags`([object_types, as_dataframe])	Get a list of all tags from skpro.
`all_objects`([object_types, filter_tags, ...])	Get a list of all objects from skpro.

General tags, packaging#

This section lists tags that are general and apply to all objects in skpro. These tags are typically used for typing, packaging and documentation purposes.

`object_type`	Type of object, e.g., 'regressor', 'transformer'.
`estimator_type`	Type of estimator, e.g., 'regressor', 'transformer'.
`reserved_params`	List of reserved parameter names.
`maintainers`	List of current maintainers of the object.
`authors`	List of authors of the object.
`python_version`	Python version specifier (PEP 440).
`python_dependencies`	Python dependencies of estimator.
`python_dependencies_alias`	Import name aliases for dependencies.
`license_type`	License type for interfaced packages.

Probabilistic regressor tags#

This section lists tags applying to probabilistic regressors ("regressor_proba" type). These tags are used to describe capabilities, properties, and behavior of probabilistic regressors.

`capability__survival`	Capability for survival analysis.
`capability__multioutput`	Support for multioutput regression.
`capability__missing`	Support for missing values.
`capability__update`	Support for online updates.
`X_inner_mtype`	Internal X machine type.
`y_inner_mtype`	Internal y machine type.
`C_inner_mtype`	Internal censoring machine type.

Distribution tags#

This section lists tags applying to probability distributions ("distribution" type). These tags are used to describe capabilities, properties, and behavior of distributions.

`capabilities__approx`	Approximate methods of distribution.
`capabilities__exact`	Numerically exact methods of distribution.
`capabilities__undefined`	Methods of distr that are mathematically undefined.
`distr__measuretype`	Measure type of distribution.
`distr__paramtype`	Parametrization type of distribution.
`approx_mean_spl`	Sample size for MC mean estimates.
`approx_var_spl`	Sample size for MC variance estimates.
`approx_energy_spl`	Sample size for MC energy estimates.
`approx_spl`	Sample size for other MC estimates.
`bisect_iter`	Max iterations for bisection method.
`broadcast_params`	Parameters to broadcast.
`broadcast_init`	Whether to initialize broadcast parameters.
`broadcast_inner`	Inner logic vectorization type.

Metric tags#

This section lists tags applying to probabilistic metrics ("metric" type).

`scitype__y_pred`	Expected input type for y_pred.
`lower_is_better`	Direction of metric optimization.
`capability__survival`	Capability for survival analysis.

Meta-object tags#

Tags relating to meta-object composition (pipelines, ensembles, etc.).

`named_object_parameters`	Component list attribute name.
`fitted_named_object_parameters`	Fitted component list attribute name.

Testing and CI tags#

These tags control behaviour of objects in skpro continuous integration tests.

They are primarily useful for developers managing CI behaviour of individual objects.

`tests__libs`	Library dependencies required for tests.
`tests__vm`	Whether tests require their own VM.
`tests__skip_by_name`	Test names to skip on CI.
`tests__python_dependencies`	Additional dependencies for tests.