Skip to main content

Table 1 Comparison of performance measures

From: Accurate and efficient gp120 V3 loop structure based models for the determination of HIV-1 co-receptor usage

Method

ACC

Se(DM)

Sp(DM)

PPV(DM)

AUC

MCC

BER

RF (LOOCV)

0.96

0.82

0.99

0.95

0.97

0.86

0.09

RF (10-fold CV) a

0.96

0.82

0.99

0.95

0.97

0.87

0.09

RF (66/34 split) a

0.96

0.84

0.99

0.94

0.97

0.87

0.08

SVM (LOOCV)

0.95

0.84

0.97

0.87

0.95

0.82

0.09

SVM (10-fold CV) a

0.95

0.83

0.97

0.86

0.95

0.81

0.10

SVM (66/34 split) a

0.95

0.85

0.97

0.84

0.96

0.81

0.09

BDT (LOOCV)

0.96

0.82

0.99

0.97

0.97

0.87

0.09

BDT (10-fold CV) a

0.96

0.80

0.99

0.96

0.97

0.85

0.10

BDT (66/34 split) a

0.97

0.83

1.00

0.98

0.97

0.89

0.08

NN (LOOCV)

0.95

0.80

0.98

0.87

0.95

0.80

0.11

NN (10-fold CV) a

0.95

0.80

0.98

0.86

0.95

0.80

0.11

NN (66/34 split) a

0.94

0.82

0.97

0.83

0.95

0.79

0.11

Sander et al.

(SVM, 10-fold CV) a

0.92

0.80

0.95

0.81

0.93

-----

-----

Prosperi et al.

(RF, 10-fold CV) a

0.88

0.63

-----

-----

0.88

-----

-----

Prosperi et al.

(SVM, 10-fold CV) a

0.90

0.69

-----

-----

0.91

-----

-----

Sing et al.

(SVM, 10-fold CV) a

-----

0.76

0.93

-----

-----

-----

-----

Xu et al.

(RF, 56/42 split) b

0.95

0.85

0.98

0.99

-----

0.87

-----

Pillai et al.

(SVM, 10-fold CV) c

0.91

0.76

0.98

0.95

-----

-----

-----

Resch et al.

(NN, 50/50 split) c

-----

0.75

0.94

0.69

-----

-----

-----

  1. aaverage over 10 iterations; bsingle split - use of duplicates inflates performance; caverage over 100 iterations