Skip to main content

Table 4 Ratios of mnoise/mtrue for each dataset and algorithm

From: The effect of noise on the predictive limit of QSAR models

Dataset/algorithm

Ridge

kNN

SVR

RF

G_298_atom

11 ± 1.3

8.8 ± 1.4

8.9 ± 0.40

8.8 ± 1.6

Alpha

6.6 ± 1.1

8.3 ± 1.3

7.3 ± 1.0

8.1 ± 1.3

Lip

20 ± 12

17 ± 9.0

7.1 ± 1.9

13 ± 5.7

Solv

5.8 ± 1.2

3.0 ± 0.23

3.3 ± 0.26

6.0 ± 0.84

BACE

13 ± 7.7

10 ± 2.8

2.9 ± 0.44

11 ± 3.0

Tox_102

44 ± 3.1

9.2 ± 2.7

–a

43 ± 3.0

Tox_134

52* ± 3.1b

14 ± 0.84

55 ± 3.3

–a

LD50

–a

9.7 ± 4.4

5.8 ± 1.2

15 ± 6.3

  1. aEntries marked with a—had a null or negative denominator
  2. bEntries marked with an * are not statistically significant