classification

No-reset classification metrics (TP, FP, FN, precision, recall, F1, report).

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetTotalTPMetric(*, error_margin, policy)

Bases: NoResetMultipleRunMetric[StateT, ProviderT, int], Generic

No-reset total true positive metric.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Bisegment policy defining true-region detection rules.

__init__(*, error_margin, policy)

Parameters:

• error_margin (tuple[int, int])

• policy (BisegmentPolicyBase[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetTotalFPMetric(*, error_margin, policy)

Bases: NoResetMultipleRunMetric[StateT, ProviderT, int], Generic

No-reset total false positive metric.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Bisegment policy defining false-region detection rules.

__init__(*, error_margin, policy)

Parameters:

• error_margin (tuple[int, int])

• policy (BisegmentPolicyBase[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetTotalFNMetric(*, error_margin, policy)

Bases: NoResetMultipleRunMetric[StateT, ProviderT, int], Generic

No-reset total false negative metric.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Bisegment policy defining true-region detection rules.

__init__(*, error_margin, policy)

Parameters:

• error_margin (tuple[int, int])

• policy (BisegmentPolicyBase[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetPrecisionMetric(*, error_margin, tp_policy, fp_policy)

Bases: NoResetDerivedMetric[StateT, ProviderT, Number, float], Generic

No-reset precision metric with independently configurable TP/FP policies.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• tp_policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Policy used for the true-positive source metric.

• fp_policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Policy used for the false-positive source metric.

__init__(*, error_margin, tp_policy, fp_policy)

Parameters:

• error_margin (tuple[int, int])

• tp_policy (BisegmentPolicyBase[StateT, ProviderT])

• fp_policy (BisegmentPolicyBase[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetRecallMetric(*, error_margin, tp_policy, fn_policy)

Bases: NoResetDerivedMetric[StateT, ProviderT, Number, float], Generic

No-reset recall metric with independently configurable TP/FN policies.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• tp_policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Policy used for the true-positive source metric.

• fn_policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Policy used for the false-negative source metric.

__init__(*, error_margin, tp_policy, fn_policy)

Parameters:

• error_margin (tuple[int, int])

• tp_policy (BisegmentPolicyBase[StateT, ProviderT])

• fn_policy (BisegmentPolicyBase[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetF1Metric(*, error_margin, precision_metric, recall_metric)

Bases: NoResetDerivedMetric[StateT, ProviderT, float, float], Generic

No-reset F1 metric derived from no-reset precision and recall metrics.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• precision_metric (NoResetPrecisionMetric[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Pre-configured no-reset precision metric.

• recall_metric (NoResetRecallMetric[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Pre-configured no-reset recall metric.

__init__(*, error_margin, precision_metric, recall_metric)

Parameters:

• error_margin (tuple[int, int])

• precision_metric (NoResetPrecisionMetric[StateT, ProviderT])

• recall_metric (NoResetRecallMetric[StateT, ProviderT])

Return type:

None

class pysatl_cpd.benchmark.online.noreset.metrics.classification.NoResetClassificationReport(*, error_margin, global_policy, precision_policy=None, recall_policy=None)

Bases: NoResetDerivedMetric[StateT, ProviderT, Number, dict[str, Number]], Generic

No-reset classification report with one global policy and optional policy overrides.

Parameters:

• error_margin (tuple[int, int]) – (Left, right) tolerance around the true change point.

• global_policy (BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]) – Default policy applied to all source metrics.

• precision_policy (Optional[BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]]) – Optional override policy for precision and its TP/FP bases.

• recall_policy (Optional[BisegmentPolicyBase[TypeVar(StateT, bound= OnlineAlgorithmState), TypeVar(ProviderT, bound= LabeledData[Any, Any])]]) – Optional override policy for recall and its TP/FN bases.

__init__(*, error_margin, global_policy, precision_policy=None, recall_policy=None)

Parameters:

• error_margin (tuple[int, int])

• global_policy (BisegmentPolicyBase[StateT, ProviderT])

• precision_policy (BisegmentPolicyBase[StateT, ProviderT] | None)

• recall_policy (BisegmentPolicyBase[StateT, ProviderT] | None)

Return type:

None