Comparative study of classification algorithms for immunosignaturing data

BMC Bioinformatics

Table 6 Performance measures of data mining algorithm at different levels of significance on A & B conditions

SIGNIFICANCE	p < 5 x 10^-4				p < 5 x 10^-3				p < 5 x 10^-2
Algorithm	Acc.	Sp	Sn	AUC	Acc.	Sp	Sn	AUC	Acc.	Sp	Sn	AUC	Avg.
Naïve Bayes	87.5	83.3	91.7	0.84	91.7	83.3	100	0.97	91.7	83.3	100	0.96	90.8
VFI	79.2	75.0	83.3	0.93	91.7	83.3	100	0.95	87.5	75.0	100	0.90	87.7
K means	87.5	83.3	91.7	0.88	91.7	83.3	100	0.92	83.3	75.0	91.7	0.83	87.5
SVM	83.3	83.3	83.3	0.83	87.5	91.7	83.3	0.87	87.5	83.3	91.7	0.88	86.1
MLP	79.2	83.3	75.0	0.70	91.7	91.7	91.7	0.95	dnf	dnf	dnf	dnf	84.7*
Hyper Pipes	83.3	75.0	91.7	0.91	83.3	83.3	83.3	0.93	70.8	83.3	58.3	0.88	82.0
Logistic R.	66.7	83.3	50.0	0.76	95.8	91.7	100	0.92	79.2	83.3	75.0	0.85	81.5
Random Forest	79.2	83.3	75.0	0.91	79.2	75.0	83.3	0.86	79.2	75.0	83.3	0.78	80.6
Bayes Net	83.3	75.0	91.7	0.87	83.3	83.3	83.3	0.83	75.0	75.0	75.0	0.67	80.2
KNN	75.0	83.3	66.7	0.85	75.0	91.7	58.3	0.90	75.0	91.7	58.3	0.84	77.8
M5P	75.0	83.3	66.7	0.74	75.0	75.0	75.0	0.79	75.0	75.0	75.0	0.74	75.2
ASC	62.5	66.7	58.3	0.65	79.2	83.3	75.0	0.85	70.8	75.0	66.7	0.76	72.0
J48	62.5	66.7	58.3	0.65	79.2	83.3	75.0	0.85	66.7	75.0	58.3	0.72	70.6
Random Tree	70.8	75.0	66.7	0.70	70.8	75.0	66.7	0.70	66.7	66.7	66.7	0.67	69.3
SLR	70.8	75.0	66.7	0.80	66.7	75.0	58.3	0.77	50.0	50.0	50.0	0.60	65.0
K star	66.7	91.7	41.7	0.83	58.3	100	46.7	0.83	50.0	0.0	100	0.50	64.3
LDA	79.2	83.3	75.0	0.84	61.2	64.5	54.5	0.52	29.2	14.3	100	0.56	62.8

Acc: Accuracy, Sp: Specificity, Sn: Sensitivity, AUC: Area under ROC curve, Avg: Average score in % for each algorithms, dnf: Did not Finish”, * denotes Avg. from 3 significance levels. Measures >90% are marked in bold.

ISSN: 1471-2105