|
4.2 Comparison of New Wave Methods for Nonparametric Multivariate Regression
David L. Banks, Mark Levenson This workshop was chosen in competition for presentation at the KDD-98 annual meeting in New York, August 27-30. The workshop discussed, in the context of data mining, various statistical techniques that attempt to avoid or minimize the Curse of Dimensionality. The lectures reviewed multivariate regression, then moved on to describe Loess, ACE, AVAS, MARS, Projection Pursuit Regression, CART, neural nets, and additive models. Examples were given from NIST data, and results of a comparative simulation study of the methods were presented.
Date created: 7/20/2001 |