Plain accuracy metric calculated based on the actual classes vs predicted classes using a default class division threshold.

AUC Score

Calculates the Area Under Curve for given labels and predictions.

Confusion Matrix

Calculates Confusion Metric on Binary Classification or Multivariate Classification labels.

F1 Score

F1 score is calculated by using the precision and recall on the label and predictions.


GINI metric is two times the ROC AUC score minus 1. This metric gives a good measure of the model’s performance when the target classes are not balanced.