Table 2. Statistical properties of neurotoxicity models against mice.
№ |
Method |
n |
m |
r2 |
s |
q2 |
scv |
Descriptors* |
Y-outliers** |
7 |
MLR |
47 |
4 |
0.595 |
0.35 |
0.370 |
0.44 |
α, ∑(Q+), ∑(Ca), m |
- |
8 |
RF |
47 |
2 |
0.786 |
0.25 |
0.536 |
0.37 |
max(Ca), ∑(Ca) |
- |
9 |
SVM |
47 |
3 |
0.569 |
0.36 |
0.435 |
0.41 |
max(Ca), ∑(Ca), ∑(Cad) |
- |
10 |
MLR |
45 |
4 |
0.741 |
0.28 |
0.670 |
0.32 |
α, ∑(Q+), ∑(Ca), m |
21, 23 |
11 |
RF |
45 |
2 |
0.835 |
0.22 |
0.633 |
0.32 |
max(Ca), ∑(Ca) |
10, 23 |
12 |
SVM |
45 |
3 |
0.623 |
0.31 |
0.491 |
0.36 |
max(Ca), ∑(Ca), ∑(Cad) |
1, 23 |
Note. * α – molecular polarizability, Å3; ∑(Q+) – sum of positive atom charges; ∑(Ca) – sum of free energy H-acceptor descriptors; μ - dipole moment, D; max(Ca) – max free energy H-acceptor descriptor; ∑(Cad) – sum of free energy H-acceptor and H-donor descriptors; ** 1: n-Pentane; 3: 10-Acetone; 21: Carbon disulfide; 23: Tetrahydrofuran