Skip to main content
Part of the Staging family.
Cleaned lap-level data from FastF1 race sessions. One row per driver × lap. Renames Bronze columns to snake_case, casts nanosecond timings to seconds, and derives the validity / track-status flags consumed across the warehouse. is_valid_lap = TRUE excludes pit laps, deleted laps, SC/VSC laps, and lap 1. Known bronze defect, not corrected here (F18, one race): 2020_1 (Austrian GP) is numbered 1-68 in bronze laps while the race ran 71. Race control puts the CHEQUERED flag on lap 71 and race_scheduled_laps says 71; on the 2026-09-25 dev build it is the only race with a chequered flag on record whose laps end short of it. The audit also found Jolpica’s pit laps for that race running 4 to 5 laps ahead of stg_pits (not re-run since). The two other feeds are ingested separately from the lap feed and neither passes through stg_laps, so the fault is in bronze ingestion, outside this layer. Effect, that race only: lap_number runs behind the true lap (3 laps at the flag, 4-5 by the pit-lap comparison), so the lap_number feature reads early and fuel_mass_kg, which counts down from lap_number at the scheduled 71-lap rate (about 1.55 kg per lap), reads about 1.5 kg per lap of offset too heavy (4.6 kg at 3 laps). The fix is a re-pull of 2020_1.

Lineage

Overview

26 tests guard this model - 15 generic, 11 singular.

Columns