Predictive classification pipeline | Data are partitioned into a training set and a testing set (75% and 25% of the samples respectively). |
Predictive classification pipeline | For each training set , the number of ranked features no that provides the smallest average KCCC is found by a nested 10X random subsampling CV. |
Predictive classification pipeline | Later, the features are ranked using the entire training set and the model is trained using the top ranked no features. |