Developer Guide for Intel® Data Analytics Acceleration Library 2019 Update 2
For two classes C1 and C2, given a vector X= (x1, …, xn) of class labels computed at the prediction stage of the classification algorithm and a vector Y= (y1, …, yn) of expected class labels, the problem is to evaluate the classifier by computing the confusion matrix and connected quality metrics: precision, recall, and so on.
QualityMetricsId for binary classification is confusionMatrix.
For description of the default quality metrics for binary classification, refer to Details.