Figure 3.

Classification performances of the control analysis

(A–C) This figure shows classification performances (i.e., accuracy) of the control analysis using behavioral data of test tasks during phases 1 (A), 2 (B), and both (C) with different feature-selecting methods (stepwise feature [➀], significant between-genotype parameters [➁ and ➂ for App^{NL-G-F/NL-G-F} > WT and App^{NL-G-F/NL-G-F} < WT, respectively], and all parameters [➃]). Asterisks (∗) denote the optimum cross-validated accuracy for the stepwise selection method. Error bars represent standard deviations of cross-validated accuracies.