estimator

Protocol definition for CUSUM parameter estimating schemas.

This module defines IEstimatingSchema, the interface for estimation components that are trained on a learning sample and optionally updated online.

class pysatl_cpd.algorithms.online.cusum.abstracts.estimator.ISchemaEstimates[source]: Bases: TypedDict

class pysatl_cpd.algorithms.online.cusum.abstracts.estimator.IEstimatingSchema[source]

Interface for estimating schemas used by generalized CUSUM.

Implementations estimate model parameters from training data, optionally update them with new observations, and expose current estimates.

abstractmethod train(train_set)[source]

Fit estimator parameters from a training sample.

Parameters:: train_set (Sequence[TypeVar(DataT)]) – Learning sample used for initial parameter estimation.
Return type:: None

abstractmethod update(observation)[source]

Update estimator parameters with a new observation.

Parameters:: observation (TypeVar(DataT)) – New observation used for adaptive update.
Return type:: None

abstractmethod reset()[source]

Reset estimator state to initial condition.

abstract property estimates: EstimatesT

Current estimated parameters.