drop

Drop-based Bayesian change-point score function.

class pysatl_cpd.algorithms.online.bayesian.component.cpf.drop.DropCPF[source]

Bases: IBayesianCPF

Return the positive drop in maximal-run probability between steps.

Sets the previous max-run-log-probability to None (first call to calculate will return 0.0).

__init__()[source]
Return type:

None

calculate(run_length_log_posterior)[source]

Compute the positive drop in max-run probability since the last step.

Parameters:

run_length_log_posterior (GenericAlias[double]) – Log-posterior probabilities for each run length.

Returns:

Positive drop (clamped at zero) in max-run probability.

Return type:

float

clear()[source]

Reset internal state.

Return type:

None