Table 12.

Real data sets used in the experiments.

Name	Domain	# of samples	# of variables	Response type	Data type	CV design	References
Infant_Mortality	clinical	5,337	86	Death within the first year	Discrete	Holdout	Mani and Cooper (1999)
Ohsumed	Text	5,000	14,373	Relevant to neonatal diseases	Continuous	Holdout	Joachims (2002)
ACPJ_Etiology	Text	15,779	28,228	Relevant to etiology	Continuous	Holdout	Aphinyanaphongs et al. (2006)
Lymphoma	Gene Expression	227	7,399	3-year survival:dead vs. alive	Continuous	10-fold	Rosenwald et al. (2002)
Gisette	Digit recognition	7,000	5,000	4 vs. 9	Continuous	Holdout	NIPS 2003 Feature Selection Challenge Guyon et al. (2006)
Dexter	Text	600	19,999	Relevant to corporate acquisitions	Continuous	10-fold	NIPS 2003 Feature Selection Challenge Guyon et al. (2006)
Sylva	Ecology	14,394	216	Ponderosa vs. rest	Continuous	Holdout	WCCI 2006 Perf. Prediction Challenge
Ovarian_Cancer	Proteomics	216	2,190	Cancer vs. normal	Continuous	10-fold	Conrads et al. (2004)
Thrombin	Drug discovery	2,543	139,351	Binding to thrombin	Discrete	Holdout	KDD Cup 2001
Breast_Cancer	Gene Expression	286	17,816	ER+vs. ER−	Continuous	10-fold	Wang et al. (2005)
Hiva	Drug discovery	4,229	1,617	Activity to HIV AIDS infection	Discrete	Holdout	WCCI 2006 Perf.Prediction Challenge
Nova	Text	1,929	16,969	Political topics vs. religious	Discrete	Holdout	WCCI 2006 Perf. Prediction Chanllenge
Bankruptcy	Financial	7,063	147	Personal bankruptcy	Continuous	Holdout	Foster and Stine (2004)