We have some telemetry data from a Formula 1 racing car doing practice laps at Silverstone. The most prominent aspect of the data is that it falls naturally into segments: going as far as you can on the straights, slowing down for turns, and speeding up again after turns. Most data mining techniques assume the data are unchanging in character. I'll present a simple technique for finding 'surprising' points where the system's behaviour seems to change, which can be implemented to work in real time. We can then look for "coincidental" changes in several variables, and even use this to cluster variables.
Last modified: Tuesday, 08-Mar-2011 09:22:37 NZDT
This page is maintained by the seminar list administrator.