If we try a loess model, we get this:
I'm not really sure what and how much we can read into these - though people do?
For a summary of the laptimes, and the stints completed by each driver throughout the session, see: China 2015 FP1 & FP2 Session Utilisation charts.
I'm not really sure what and how much we can read into these - though people do?
For a summary of the laptimes, and the stints completed by each driver throughout the session, see: China 2015 FP1 & FP2 Session Utilisation charts.
I can do the explanation for what we're trying to read into it. Qualifying times, and short run times, tend to be done on low fuel, and therefore don't reflect race pace. So I'm mostly deeply suspicious of Rosberg's times, or I would be if he weren't in a Mercedes.
ReplyDeleteOne way to try to use data like these plots might be to "normalize" historical data for the same teams/drivers by their qualifying time from each weekend. Barring circumstances like rain or terrible traffic, a qualifying time can generally be expected to reflect a car-driver combo at its best--minimal dissembling or sandbagging. So looking at past trends for FP lap time vs qual pace may reveal trends in how much dissembling or sandbagging a team tends to engage in during The FPs.
ReplyDeleteAnother thing we can take from the long-run FP data would be to get a sense (average and standard deviation) of how much faster each car/driver gets as fuel burns. These numbers would be helpful on race day to anyone interested in whether driver A has a chance of closing the gap on driver B.
@Andy - using qualifying times to "back correct" FP times is an interesting thought (I guess by looking across drivers we might also be able to account for track evolution?) Similarly comparing race laptimes with long run practice times to try to validate any models that we can produce. (On my to do list is trying to replicate this approach - https://f1metrics.wordpress.com/2014/10/03/building-a-race-simulator/ )
ReplyDeleteWrt "get[ting] a sense (average and standard deviation) of how much faster each car/driver gets as fuel burns", what method did you have in mind for calculating that?
Additional comments from around the web:
ReplyDelete- Ash, in a comment responding to my comment linking to this post at http://www.jamesallenonf1.com/2015/04/what-can-we-learn-from-f1-practice-about-how-chinese-gp-will-turn-out/#comment-3171157