Applications of Bayesian Gene Selection and Classification with Mixtures of Generalized Singular 𝑔-Priors

Table 4

Performance comparison of different procedures for the leukemia study.

Methods	No. of genes	Testing error rate	Testing accuracy
Bayesian 𝑔-prior	6	0.0588 (2/34)	0.9412 (32/34)
Bayesian 𝑔-prior	10	0.0588 (2/34)	0.9412 (32/34)
Bayesian 𝑔-prior	14	0.0588 (2/34)	0.9412 (32/34)
Weighted voting machine^a	50	0.1471 (5/34)	0.8529 (29/34)
MAVE-LD^b	50	0.0294 (1/34)	0.9706 (33/34)
Two-step EBM^c	32	0.1471 (5/34)	0.8529 (29/34)
Two-step EBM^c	256	0.0588 (2/34)	0.9412 (32/34)
Two-step EBM^c	512	0.0294 (1/34)	0.9706 (33/34)
KIGP + PK^d	20	0.0588 (2/34)	0.9412 (32/34)
𝑡-test + probit regression	6	0.1765 (6/34)	0.8235 (28/34)
𝑡-test + probit regression	10	0.0882 (3/34)	0.9118 (31/34)
𝑡-test + probit regression	14	0.1176 (4/34)	0.8824 (30/34)

aProposed by Gloub et al. [36].
^bProposed by Antoniadis et al. [43].
^cProposed by Ji et al. [47].
^dProposed by Zhao and Cheung [48].