Skip to main content
Part of the Skill family.
Driver isolation step 0: the clean-lap panel Ω and the field-centred isolated time. One row per Ω lap: correction_weight 1, anomaly_class normal/clean_cliff, dry, not rain, not an out- or in-lap, a known tyre (compound_component_s and laps_past_cliff non-NULL, never COALESCEd), not in var(‘isolation_exclude_races’), on a (race, lap) with >= var(‘isolation_min_field_n’) such laps. y_s = x_s minus the lap’s field median, so every per-(race, lap) term (field base, fuel, rubber, ambient) cancels: the ratings do not depend on the FD1 base (T41). Also the stint phase (tyre phase from the seed’s onset, recovery overlay from the air state). The input of the car-term fit (fit_constructor_car_fe.py —panel isolation). Leakage: a function of the residual trajectory; no ML-contract mart may depend on it (T48, ml/tests/test_manifest_contract.py). WI-16a, _roadmap/_fixes/wi/WI-16-cumulative- driver-isolation.md.

Lineage

Overview

31 tests guard this model - 27 generic, 4 singular.

Columns