MODULE REFERENCE

capabilities

anomx.base.capabilities

Capability declarations for pipeline components.

Two complementary mechanisms live here:

Capability mixins express *what a normality model can do* (predict, reconstruct, embed, estimate densities or boundaries). They are composed onto implementations, e.g. class TemporalFusionTransformer(NormalityModel, Predictive, Representational).
ComponentCapabilities expresses *what data a component handles*, validated against observation-set characteristics before execution.

BoundaryEstimating

BoundaryEstimating

Bases: Capability

The model can estimate a decision boundary around normal data.

Members

capability_keyattribute

# BoundaryEstimating.capability_key Source ↗

capability_key = 'boundary_estimating'

decision_functionmethod

# BoundaryEstimating.decision_function Source ↗

decision_function(data: Any) -> Any

Return signed distances to the normality boundary.

Signature parameters
Parameter	Type	Default
`data`	`Any`	`required`

Returns Any

Inherited members

get_capability_keyCapability

classView source ↗

Capability

Capability

Marker base for model capability mixins.

Members

capability_keyattribute

# Capability.capability_key Source ↗

capability_key: ClassVar[str] = ''

get_capability_keyclassmethod

# Capability.get_capability_key Source ↗

get_capability_key() -> str

Returns str

classView source ↗

ComponentCapabilities

ComponentCapabilities(supports_unordered_samples: bool = True, supports_sequences: bool = False, supports_streaming: bool = False, supports_multivariate: bool = True, supports_variable_length: bool = True, supports_irregular_sampling: bool = True, supports_missing_values: bool = False, supports_graphs: bool = False, supports_hierarchies: bool = False, requires_regular_sampling: bool = False, requires_fixed_length: bool = False, requires_training: bool = True, requires_labels: bool = False, batch_fit: bool = True, incremental_fit: bool = False, batch_inference: bool = True, streaming_inference: bool = False)

What a component supports and what it requires from its input.

Signature parameters
Parameter	Type	Default
`supports_unordered_samples`	`bool`	`True`
`supports_sequences`	`bool`	`False`
`supports_streaming`	`bool`	`False`
`supports_multivariate`	`bool`	`True`
`supports_variable_length`	`bool`	`True`
`supports_irregular_sampling`	`bool`	`True`
`supports_missing_values`	`bool`	`False`
`supports_graphs`	`bool`	`False`
`supports_hierarchies`	`bool`	`False`
`requires_regular_sampling`	`bool`	`False`
`requires_fixed_length`	`bool`	`False`
`requires_training`	`bool`	`True`
`requires_labels`	`bool`	`False`
`batch_fit`	`bool`	`True`
`incremental_fit`	`bool`	`False`
`batch_inference`	`bool`	`True`
`streaming_inference`	`bool`	`False`

Members

supports_unordered_samplesattribute

# ComponentCapabilities.supports_unordered_samples Source ↗

supports_unordered_samples: bool = True

supports_sequencesattribute

# ComponentCapabilities.supports_sequences Source ↗

supports_sequences: bool = False

supports_streamingattribute

# ComponentCapabilities.supports_streaming Source ↗

supports_streaming: bool = False

supports_multivariateattribute

# ComponentCapabilities.supports_multivariate Source ↗

supports_multivariate: bool = True

supports_variable_lengthattribute

# ComponentCapabilities.supports_variable_length Source ↗

supports_variable_length: bool = True

supports_irregular_samplingattribute

# ComponentCapabilities.supports_irregular_sampling Source ↗

supports_irregular_sampling: bool = True

supports_missing_valuesattribute

# ComponentCapabilities.supports_missing_values Source ↗

supports_missing_values: bool = False

supports_graphsattribute

# ComponentCapabilities.supports_graphs Source ↗

supports_graphs: bool = False

supports_hierarchiesattribute

# ComponentCapabilities.supports_hierarchies Source ↗

supports_hierarchies: bool = False

requires_regular_samplingattribute

# ComponentCapabilities.requires_regular_sampling Source ↗

requires_regular_sampling: bool = False

requires_fixed_lengthattribute

# ComponentCapabilities.requires_fixed_length Source ↗

requires_fixed_length: bool = False

requires_trainingattribute

# ComponentCapabilities.requires_training Source ↗

requires_training: bool = True

requires_labelsattribute

# ComponentCapabilities.requires_labels Source ↗

requires_labels: bool = False

batch_fitattribute

# ComponentCapabilities.batch_fit Source ↗

batch_fit: bool = True

incremental_fitattribute

# ComponentCapabilities.incremental_fit Source ↗

incremental_fit: bool = False

batch_inferenceattribute

# ComponentCapabilities.batch_inference Source ↗

batch_inference: bool = True

streaming_inferenceattribute

# ComponentCapabilities.streaming_inference Source ↗

streaming_inference: bool = False

validate_againstmethod

# ComponentCapabilities.validate_against Source ↗

validate_against(characteristics: DataCharacteristics) -> list[str]

Return human-readable problems, empty when the pipeline is valid.

Signature parameters
Parameter	Type	Default
`characteristics`	`DataCharacteristics`	`required`

Returns list[str]

to_dictmethod

# ComponentCapabilities.to_dict Source ↗

to_dict() -> dict[str, Any]

Returns dict[str, Any]

classView source ↗

Distributional

Distributional

Bases: Capability

The model can estimate densities or likelihoods.

Members

capability_keyattribute

# Distributional.capability_key Source ↗

capability_key = 'distributional'

log_likelihoodmethod

# Distributional.log_likelihood Source ↗

log_likelihood(data: Any) -> Any

Return per-observation log likelihoods.

Signature parameters
Parameter	Type	Default
`data`	`Any`	`required`

Returns Any

Inherited members

get_capability_keyCapability

classView source ↗

Predictive

Predictive

Bases: Capability

The model can forecast expected future values.

Members

capability_keyattribute

# Predictive.capability_key Source ↗

capability_key = 'predictive'

forecastmethod

# Predictive.forecast Source ↗

forecast(data: Any, horizon: int) -> Any

Forecast horizon future steps.

Signature parameters
Parameter	Type	Default
`data`	`Any`	`required`
`horizon`	`int`	`required`

Returns Any

Inherited members

get_capability_keyCapability

classView source ↗

Reconstructive

Reconstructive

Bases: Capability

The model can reconstruct its input from a compressed representation.

Members

capability_keyattribute

# Reconstructive.capability_key Source ↗

capability_key = 'reconstructive'

reconstructmethod

# Reconstructive.reconstruct Source ↗

reconstruct(data: Any) -> Any

Return the reconstruction of the input.

Signature parameters
Parameter	Type	Default
`data`	`Any`	`required`

Returns Any

Inherited members

get_capability_keyCapability

classView source ↗

Representational

Representational

Bases: Capability

The model can embed observations into a latent representation.

Members

capability_keyattribute

# Representational.capability_key Source ↗

capability_key = 'representational'

embedmethod

# Representational.embed Source ↗

embed(data: Any) -> Any

Return latent embeddings for the input.

Signature parameters
Parameter	Type	Default
`data`	`Any`	`required`

Returns Any

Inherited members

get_capability_keyCapability

functionView source ↗

collect_capability_keys

collect_capability_keys(component_class: type) -> list[str]

Collect the capability keys mixed into a component class.

Signature parameters
Parameter	Type	Default
`component_class`	`type`	`required`

Returns list[str]