fig7
From: Intelligent visualization-driven materials design via two-dimensional symbolic feature generation
Figure 7. Comparison between multiple dimensionality reduction methods and the 2D-SFG method. (A) Thermal hysteresis of shape memory alloys; (B) Hardness and electric conductivity of copper alloys. Error bars indicate the standard deviations obtained from 10-fold cross-validation. 2D-SFG: Two-dimensional symbolic feature generation; RMSE: root mean square error; PCA: principal component analysis; KPCA: kernel principal component analysis; LLE: locally linear embedding; MDS: multidimensional scaling; MAPE: mean absolute percentage error.
