Skip to main content
. 2021 Mar 5;11:611437. doi: 10.3389/fonc.2021.611437

Table 5.

Performance metrics for voting model, calculated over 1,000 test iterations.

Performance Metric
Accuracy 91.5 ± 0.9%
Precision 87.3 ± 1.1%
Recall 94.5 ± 1.7%
F1 Score 90.6 ± 1.1%
ROC-AUC 91.8 ± 1.0%
Leave-One-Out Accuracy (LOOCV) 89.80%

For each iteration, the KFold random state was set to the iteration number. The LOOCV accuracy is not calculated over these 1,000 test iterations as every data point acts as the test set once, meaning there are no random seeds that can be implemented.