Table 2 Results of k-fold, \(K=5\), grid search cross validation for random forest classifier using multiple metric evaluation.

From: Selecting the most important self-assessed features for predicting conversion to mild cognitive impairment with random forest and permutation-based methods

 

Training

Test

 

AUC

Precision

Accuracy

AUC

Precision

Accuracy

AUC

0.89

0.829

0.8478

0.54

0.25

0.848

Precision

0.77

0.391

0.8315

0.56

0.187

0.77

Accuracy

0.98

0.97

0.992

0.52

0.33

0.875

  1. Each estimator is refitted using the best combination of hyperparameters. Random Forest is refitted using three scorers -AUC, precision, and accuracy- for either train and the test sets. In detail description of random forest fitting is given in the Supplementary Material.
Back to article page