(a) (b)

The ROC curve of LDA model for the seeds data using pROC. The shaded areas

tand for the confidence intervals, which are generated using the bootstrapping

a) The full ROC analysis. (b) The partial ROC analysis.

(a) (b)

he comparison of AUC distributions of two sets of the cross-validation models

for the seeds data. The legends show the standard deviation of 1,000 AUC

ach panel. (a) The distribution of the AUCs of the models constructed using the

ss-validation approach. (b) The distribution of the AUCs of the models

using the 10-fold cross-validation.

monstrate that the Jackknife test is robust, a comparison between

of K-fold cross-validation models was used here. One set of

were constructed by using the 3-fold cross-validation approach. It

ated for 1,000 times. The other set of models were constructed by

10-fold cross-validation approach. It was also repeated for 1,000

oth sets were based on the seeds data. Each classifier using either

oss-validation or 10-fold cross-validation was evaluated using the

lysis. It can be seen that when the fold number was increased, the