annotations

Annotation type definitions for time series data.

type pysatl_cpd.data.typedefs.annotations.AnnotationBuilder = Callable[[Sequence], MergedAnnotationT]

class pysatl_cpd.data.typedefs.annotations.TimeseriesAnnotation(*, name, source=None, metadata=<factory>)[source]

Bases: object

Base annotation for time series data.

This class is the common source for all time series annotations, providing a name, optional source, and metadata storage.

Variables:

name – Unique identifier for this annotation.
source – Optional source identifier or description.
metadata – Additional key-value metadata storage.

Parameters:

name (str)
source (str | None)
metadata (frozendict[str, Hashable])

name: str

source: str | None

metadata: frozendict[str, Hashable]

property provider_type: ProviderType

Returns:: The provider type identifier for time series annotations.
Return type:: ProviderType

__hash__()[source]

Return a stable hash for the annotation identity.

Return type:: int

__init__(*, name, source=None, metadata=<factory>)

Parameters:

name (str)
source (str | None)
metadata (frozendict[str, Hashable])

Return type:

None

class pysatl_cpd.data.typedefs.annotations.UnlabeledTimeseriesAnnotation(*, name, source=None, metadata=<factory>)[source]

Bases: TimeseriesAnnotation

Annotation for unlabeled time series data.

This annotation type represents time series without explicit segment or change point labels, used for unsupervised detection.

Parameters:

name (str) – Unique identifier for this annotation.
source (str | None) – Optional source identifier or description.
metadata (frozendict[str, Hashable]) – Additional key-value metadata storage.

property provider_type: ProviderType

Returns:: The provider type identifier for unlabeled time series.
Return type:: ProviderType

__hash__()[source]

Return a stable hash for unlabeled annotations.

Return type:: int

__init__(*, name, source=None, metadata=<factory>)

Parameters:

name (str)
source (str | None)
metadata (frozendict[str, Hashable])

Return type:

None

class pysatl_cpd.data.typedefs.annotations.SegmentAnnotation(*, name, source=None, metadata=<factory>, state)[source]

Bases: TimeseriesAnnotation

Annotation with segment state information.

This annotation type represents time series with a segment state descriptor, used for supervised change point detection.

Variables:

state – Descriptor for the segment state.

Parameters:

name (str)
source (str | None)
metadata (frozendict[str, Hashable])
state (StateDescriptor)

state: StateDescriptor

property provider_type: ProviderType

Returns:: The provider type identifier for segment annotations.
Return type:: ProviderType

__hash__()[source]

Return a stable hash for segment annotations.

Return type:: int

__init__(*, name, source=None, metadata=<factory>, state)

Parameters:

name (str)
source (str | None)
metadata (frozendict[str, Hashable])
state (StateDescriptor)

Return type:

None

class pysatl_cpd.data.typedefs.annotations.NoChangeSeriesAnnotation(*, name, source=None, metadata=<factory>, state)[source]

Bases: SegmentAnnotation

Annotation for series without change points.

This annotation type represents a time series where no change points are expected, used as negative examples in training.

Parameters:

name (str) – Unique identifier for this annotation.
source (str | None) – Optional source identifier or description.
metadata (frozendict[str, Hashable]) – Additional key-value metadata storage.
state (StateDescriptor) – Descriptor for the segment state.