CATS
84 Citations•2005•
N. Serban, L. Wasserman
Journal of the American Statistical Association
CATS—clustering after transformation and smoothing—is a technique for nonparametrically estimating and clustering a large number of curves for screening out flat curves, clustering curves with similar shape, and nonparametrically inferring the clustering estimation error rate.
Abstract
CATS—clustering after transformation and smoothing—is a technique for nonparametrically estimating and clustering a large number of curves. Our motivating example is a genetic microarray experiment, but the method is very general. The method includes transformation and smoothing multiple curves, multiple nonparametric testing for screening out flat curves, clustering curves with similar shape, and nonparametrically inferring the clustering estimation error rate.