Iterative Reweighted Noninteger Norm Regularizing SVM for Gene Expression Data Classification

Table 1

(a) Comparison of the prediction error (Pe%) and feature selection error (Fse%) on simulated datasets for different 𝑝 values. (b) Comparison of variable selection and the prediction performance on simulated datasets for different 𝑝 values.

(a)  

Training + validation	0.1	0.2	0.3	0.4	0.5
250 + 750
Pe	0.47	0.43	0.45	0.42	0.38
Fse	8.00	8.33	6.5	6.83	5.33
500 + 500
Pe	0.37	0.34	0.32	0.31	0.29
Fse	2.67	2.67	1.67	1.17	0.50
750 + 250
Pe	0.35	0.30	0.32	0.34	0.31
Fse	0.50	0.17	0.33	0	0

(b)  

Training + validation	0.6	0.7	0.8	0.9	1.0	2.0
250 + 750
Pe	0.38	0.38	0.41	0.41	0.40	0.42
Fse	4.67	3.67	2.00	2.33	2.50	4.17
500 + 500
Pe	0.28	0.33	0.28	0.30	0.31	0.34
Fse	0.17	0	0	0	0	0.83
750 + 250
Pe	0.31	0.28	0.26	0.28	0.37	0.34
Fse	0	0.17	0	0	0	0.17