Re: Goodness of fit measures for a distribution

Unknown wrote:
So basically, based on the recommendations from this newsgroup, it is impossible to choose which is the best distribution to fit your data, considering all you know about the data is that it comes from a random process, since there are really no objective metrics and all you have to do is rely on graphic plots (which by the way dont tell much since many models will give you similar fits).

Would that be an accurate statement?

Well, it's one I would go for. It's time to trot out the well known G.E.P. Box quote: "All models are wrong, but some are useful". This suggests that "best" is not a good criterion (because it'll never be right anyway, and there are an infinite number of models you could try). I suppose the criterion used in practice is "good enough". And we try to discourage people from putting a number on that as well.

Bob

--
Bob O'Hara
Department of Mathematics and Statistics
P.O. Box 68 (Gustaf Hällströmin katu 2b)
FIN-00014 University of Helsinki
Finland

Telephone: +358-9-191 51479
Mobile: +358 50 599 0540
Fax:  +358-9-191 51400
WWW:  http://www.RNI.Helsinki.FI/~boh/
Journal of Negative Results - EEB: www.jnr-eeb.org
.

Relevant Pages