int_driver_season_ratings

One row per (driver_id, season).

Part of the Skill family.

One row per (driver_id, season). Aggregates the race-grain equal-car rating (int_driver_race_skill_loro.driver_skill_loro_s, the median lap-by-lap gap to the teammate) to season grain, shrunk toward the season mean over all drivers (Bayesian, prior = 5 virtual races). Negative = faster than his teammate(s). Teammate-relative, so a car-era shift cancels inside each pair. Intermediate for int_era_normalized_driver_rating.

​

Lineage

​

Overview

Property	Value
Layer	intermediate
Family	Skill
Contract enforced	No
Tags	driver_rating, era_rating
Upstream	int_driver_race_skill_loro

14 tests guard this model - 12 generic, 2 singular.

​

Columns

Column	Type	Nullable	Tests	Description
driver_season_id		Yes	not_null, unique	Surrogate PK = hash(driver_id, season)
driver_id		Yes	not_null, relationships	Driver identifier
season		Yes	not_null	Race year (integer)
n_races		Yes	not_null	Races contributing to this estimate
raw_residual_mean_s		Yes	not_null	Unshrunk season mean gap to the teammate. Negative = faster than the teammate.
shrunk_residual_s		Yes	not_null	Bayesian posterior mean toward the season mean over all drivers.
shrunk_residual_se_s		Yes	not_null, expect_column_values_to_be_between	Posterior standard error in seconds.
rating_confidence		Yes	not_null, expect_column_values_to_be_between	Data fraction of posterior mass. n_races / (n_races + 5). [0, 1].