constant

Constant hazard model for Bayesian online change-point detection.

class pysatl_cpd.algorithms.online.bayesian.component.hazard.constant.ConstantHazard(lambda_)[source]

Bases: IHazard

Constant hazard model with fixed timescale.

Parameters:: lambda_ (float) – Expected run length (must be >= 1.0).
Raises:: ValueError – If lambda_ < 1.0.

__init__(lambda_)[source]

Parameters:: lambda_ (float)
Return type:: None

hazard(run_lengths)[source]

Return constant log-hazard and log-survival for each run length.

Parameters:: run_lengths (GenericAlias[int_]) – Array of run length indices (used only for shape).
Returns:: Pair of (log_hazard, log_survival) arrays broadcast to match the shape of run_lengths.
Return type:: tuple[GenericAlias[double], GenericAlias[double]]