posted on 2019-07-31, 09:17authored byLingfei Wang, Tom Michoel
Simulation 1 confirmed superior and consistent performances of dimension reduction methods, especially PCA and Isomap. (A) Comparison of dimension reduction, individual predictions, and the class probability in AUROC and AUPR (cf Figure 3A) for simulation 1 (Table S2). (B) Comparison of dimension reduction and supervised learning in cross validation at 25% training data (cf Figure 4) for simulation 1. Color reflects relative ranking. (C, D, E, F) ROC (C, E) and Precision-Recall (D, F) curves for dimension reductions, existing crowd wisdoms, the class probability, and individual predictions of simulation 1. In C, D, the best parameter (in A) was selected according to AUROC (C) or AUPR (D) for each parametric dimension reduction and PCA was selected for non-parametric dimension reduction. All methods are shown in E, F. Existing crowd wisdoms were performed on binarized input data.