fig6

Figure 6. Performance of the 2D-SFG method on classification problems. (A and B) Evaluation metrics of the classification models; (C) Comparison of modeling accuracy between direct modeling and symbolic feature generation; (D) Comparison of multiple dimensionality reduction methods with the 2D-SFG method. Error bars indicate the standard deviation obtained from 10-fold cross-validation. 2D-SFG: Two-dimensional symbolic feature generation; AUC: area under the curve; PCA: principal component analysis; KPCA: kernel principal component analysis; LLE: locally linear embedding; MDS: multidimensional scaling.