Skip to main content
. 2020 Jul 17;11:673. doi: 10.3389/fpsyt.2020.00673

Figure 3.

Figure 3

Results in different combinations of hyper-parameters. (A) Discriminatory ability – the accuracy of the imputed dataset in classifying ADHD vs. typically-developing (TD) controls for each iteration. (B) Predictive power – the accuracy of deep neural network (DNN) to impute candidate question’s missing values for each iteration. (C) Time cost – the processing time required for imputation for each iteration.