elbow_helper.candidates module

Phase 3 — candidate generation with the from-scratch locator.

Sweeps every (smoothing window, sensitivity) setting, runs KneeLocator in online mode, and collects every returned knee as a KneeCandidate — annotated with metrics but not yet accepted.

Author

Warith Harchaoui, <warith.harchaoui@deraison.ai>

elbow_helper.candidates.generate_candidates(prepared, config)[source]

Run the full scale-space × sensitivity locator sweep.

Parameters:

prepared (PreparedCurve) – The normalized curve.
config (RobustKneeConfig) – Supplies the smoothing and sensitivity grids.

Returns:

Every knee found, annotated with metrics. No acceptance yet.

Return type:

list of KneeCandidate

elbow_helper.candidates.sensitivity_grid(n, config)[source]

Distinct sensitivities S from sensitivity_fractions.

Parameters:

n (int)
config (RobustKneeConfig)

Return type:

List[float]