Table 5 Comparison of ANN and SVR models for estimating the Gibbs energy of sublimation.

From: Predicting the Enthalpy and Gibbs Energy of Sublimation by QSPR Modeling

Equation 13 14 15 16
Artificial Neural Networks
R2 (train.) 0.80 0.59 0.71 0.73
Std. dev.a 8.9 16.6 9.4 9.6
Largest deviationa
   Positive 28.1 38.8 40.2 36.9
   Negative −31.3 −41.3 −32.8 −33.0
R2 (test) 0.63 −0.32 0.59 0.57
Std. dev.a,b 8.7 11.5 9.3 9.5
Support Vector Regression
R2 (train.) 0.77 0.51 0.70 0.71
Std. dev.a 9.2 12.4 9.8 9.6
Largest deviationa
   Positive 38.5 38.0 37.3 39.5
   Negative −36.5 −53.3 −33.0 −34.8
R2 (test) 0.61 0.26 0.56 0.58
Std. dev.a,b 9.0 11.9 9.8 9.5