Fig. 4: Input/output correlations from raw data and SHAP (Shapley value based) analyses.
From: Machine learning-assisted crystal engineering of a zeolite
a Correlation between Na2O/Al2O3 in synthesis mixture (input) and Si/Al ratio of product via ICP (output). b Correlation between Si/Al ratio via ICP (output) and FAU fraction (output), the latter refers to FAU/(FAU + EMT). c Correlation between FAU fraction (output) and particle size (output). In (a–c), blue/red/green dots represent training/testing/prediction points, respectively. d The constructed SHAP summary plot for the trained Si/Al model “rescaled LOOCV (w/o categorical)”. e The waterfall plot on the model “rescaled LOOCV (w/o categorical)” for entry 91 indicating the contribution of each variable (synthesis condition) for the predicted normalized output of the model \(f(x)\). To the estimated value of the output \(f(x)\) the denormalization \({f}_{{True}}\left(x\right)=f\left(x\right){\sigma }_{{Si}/{Al}}+{\mu }_{{Si}/{Al}}\) is needed to recover the true values, where \({\sigma }_{{Si}/{Al}}\) corresponds to the standard deviation and \({\mu }_{{Si}/{Al}}\) corresponds to the mean of the training set.
