Skip to main content

Advertisement

Table 4 Prediction of the DFT DM by random forests using NBO charges and combining different descriptors

From: Machine learning for the prediction of molecular dipole moments obtained by density functional theory

Models Training seta Test set
MAE (D) R2/RMSE (D) MAE(D) R2/RMSE (D)
Ab 0.627 0.757/0.912 0.623 0.774/0.905
Bc 0.616 0.762/0.903 0.611 0.778/0.896
Cd 0.525 0.823/0.780 0.512 0.846/0.752
De 0.522 0.824/0.777 0.509 0.846/0.750
Ef 0.562 0.790/0.850 0.553 0.807/0.838
Fg 0.497 0.837/0.749 0.479 0.860/0.719
  1. aOOB estimation
  2. bRDF pairs NBO charges, PchmDM NBO charges, and DMNBO
  3. cRDF pairs NBO charges, PchmDM NBO charges, geometric CDK, and DMNBO
  4. dRDF pairs NBO charges, PchmDM NBO charges, DMPEOE, and DMNBO
  5. eRDF pairs NBO charges, PchmDM NBO charges, geometric CDK, DMPEOE, and DMNBO
  6. fRDF pairs NBO charges, PchmDM NBO charges, MACCS fingerprints, and DMNBO
  7. gRDF pairs NBO charges, PchmDM NBO charges, MACCS fingerprints, DMPEOE, and DMNBO