Formula 1 lap time modeling using generalized additive models

In this paper we aim at describing the Formula 1 team and driver performances, during a particular race, using a suitable generalized additive model for the representation of lap time evolution. Although the modeling of Formula 1 races is an extremely complicated task, our model gives a rather flexible specification of the lap times as a function of some relevant numeric and categorical predictor variables. We fit the model using the freely available data regarding the Formula 1 season 2015 and we find that, for a grand prix without unpredictable events such as safety car, virtual safety car or race suspension, the model provides an accurate description of the race development. Moreover, it can be fruitfully considered for specifying alternative race strategies, with particular regard to the pit stop choices.