Open Access
March 2010 Feature selection in omics prediction problems using cat scores and false nondiscovery rate control
Miika Ahdesmäki, Korbinian Strimmer
Ann. Appl. Stat. 4(1): 503-519 (March 2010). DOI: 10.1214/09-AOAS277


We revisit the problem of feature selection in linear discriminant analysis (LDA), that is, when features are correlated. First, we introduce a pooled centroids formulation of the multiclass LDA predictor function, in which the relative weights of Mahalanobis-transformed predictors are given by correlation-adjusted t-scores (cat scores). Second, for feature selection we propose thresholding cat scores by controlling false nondiscovery rates (FNDR). Third, training of the classifier is based on James–Stein shrinkage estimates of correlations and variances, where regularization parameters are chosen analytically without resampling. Overall, this results in an effective and computationally inexpensive framework for high-dimensional prediction with natural feature selection. The proposed shrinkage discriminant procedures are implemented in the R package “sda” available from the R repository CRAN.


Download Citation

Miika Ahdesmäki. Korbinian Strimmer. "Feature selection in omics prediction problems using cat scores and false nondiscovery rate control." Ann. Appl. Stat. 4 (1) 503 - 519, March 2010.


Published: March 2010
First available in Project Euclid: 11 May 2010

zbMATH: 1189.62102
MathSciNet: MR2758182
Digital Object Identifier: 10.1214/09-AOAS277

Keywords: “small n, large p” setting , Correlation , correlation-adjusted t-score , False discovery rates , Feature selection , higher criticism , James–Stein estimator , linear discriminant analysis

Rights: Copyright © 2010 Institute of Mathematical Statistics

Vol.4 • No. 1 • March 2010
Back to Top