Table 1:

hyperSMURF learning hyper-parameters

Parameter	Description
nParts	Number of parts of the partition
fp	Multiplicative factor for oversampling the minority class. For instance with fp = 2 two novel examples are synthesized for each positive example of the original dataset, according to the SMOTE algorithm
ratio	Ratio for the undersampling of the majority class. For instance, ratio = 2 sets the number of negative examples as twice the total number of original and oversampled positive examples
k	Number of the nearest neighbours of the SMOTE algorithm
nTrees	Number of trees included in each random forest
mTry	Number of features to be randomly selected at each node of the decision trees included in the random forest