Skip to main content
. 2019 Jun 8;20(11):2801. doi: 10.3390/ijms20112801

Table 5.

Model performance as a function of the fingerprint types used as molecular descriptors to build the kernel-based partial least squares (KPLS) models.

Fingerprint Score q 2 r 2 RMSE SD N
Dendritic 0.76 0.82 0.89 0.40 0.53 3
Linear 0.78 0.83 0.89 0.41 0.51 3
Radial 0.80 0.81 0.80 0.54 0.54 2
Molprint2D 0.82 0.84 0.81 0.52 0.50 3

Note: q2 = correlation coefficient for the test set (r2pred); r2 = correlation coefficient for the training set; RMSE = root mean square error; SD = standard deviation; N = number of components; score = (1 − |(r2 − q2)|) × q2.