Skip to main content
Fig. 4 | Journal of Cheminformatics

Fig. 4

From: Naturally-meaningful and efficient descriptors: machine learning of material properties based on robust one-shot ab initio descriptors

Fig. 4

a The feature importance matrix for predicting the properties by the descriptor groups outlined in Section “Features”. The properties shown include the PBE bandgap, EPBE, the HSE bandgap with (\({E}_{HSE}^{PBE}\)) and without (EHSE) using the PBE bandgap as a feature, the GW bandgap with (\({E}_{GW}^{PBE}\)) and without (EGW) using the PBE bandgap as a feature, and the exciton binding energy Eb. (b, c) The correlation plots for and linear fitting for the HSE and GW bandgaps versus the PBE bandgaps. (dh) The correlation plots for the prediction of the regression models for EPBE, EHSE, EGW and Eb, respectively

Back to article page