Janet Grassmann, Martin Rezcko, Sandor Suhai and Lutz Edler
Department of Statistics, Stanford University,
Palo Alto, CA 94305 USA
Synaptic Ltd Aristotelous 313, 13671 Acharnai, Greece
Department of Molecular Biophysics and
Biostatistics Unit, German Cancer Research Center, D-96120 Heidelberg
Abstract
Feed forward neural networks are compared with standard and new
statistical classification procedures for the classification of
proteins. We applied logistic regression, an additive model and
projection pursuit regression form the methods based on a posteriori
probabilities: linear, quadratic and a flexible discriminant analysis
from the methods based on class conditional probabilities: and
the K-nearest neighbors classification rule. Both, the apparent
error rate obtained with the training sample (n=143) and the test
error rate obtained with the test sample (n=125), and the 10-fold
cross-validation error were calulated. We conclude that some of
the standard statistical methods are potent competitors to the
more flexible tools of machine learning.