Re:Richard P. Feynman - The Relation of Science and Religion
by Herring on Tuesday February 16, 2010 @10:16PM
(#1130733)
Well there's nothing really special about the "next" part. It's just that you can't check the accuracy ("truth" is a dirty word, apparently) of your model against the data you used to construct it. It's why there's stuff like cross validation in machine learning.
Anyway a model is not really useful either when there's only a finite amount of data. I hear astronomy's starting to hit against this problem.
Anyway a model is not really useful either when there's only a finite amount of data. I hear astronomy's starting to hit against this problem.