Data Stream Classification

Drift Detection Method (DDM)

Statistical technique for detecting concept drift by monitoring the classifier's error rate and its variations. It uses confidence bounds based on the binomial distribution to identify when model performance degrades significantly.

← Terug