Am Mittwoch 20 Februar 2008 schrieb Mike Schuh: > I wonder if a multiple regression against several factors would turn up > anything useful: > > vis = f(humidity,temp,wind,ceiling,elevation,sun_angle,ground_cover) > > Not sure how to incorporate sun_angle for values < 0 (i.e., night). Moon? > > ground_cover might allow for dusty areas v. urban settings, etc. > > This line of investigation sounds very interesting!
Yeah, given that the problem is multidimensional I also wonder what that gives. Note though that the effects might be (and most probably are) nonlinear, so that linear regression gives wrong results. Anyway for a quick look for any effect at all it should be ok. @Melchior: Can you please post the raw data set anywhere (or PM me). I'd like to try some math on them (PCA to look for clusters, ...). I'd appreciate pre processed data (i.e. temps, etc. already parsed from the metar reports), but a text file with raw metar reports is also ok. Just takes longer as I've to write a parser first If nothing useful comes up, as a last resort we can maybe tabulate probability density distributions from the (filtered) data and use these for random sampling. We may even mix both approaches in the form of Bayesian Trees. As a very last (but quickest) resort a conservative setting is perfectly ok :) Thomas ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Flightgear-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/flightgear-devel
