Classification learning on non-stationary data may face dynamic changes from time to time. The major problem in it is the class imbalance and high cost of labeling instances despite drifts. Imbalance is due to lower number of samples in the minority class than the majority class. Imbalanced data results in the misclassification of data points.