Ensemble for non stationary data stream: Performance improvement over learn++.NSE

Now a day, non-stationary data stream is a challenging and active topic among the researcher because the joint probability distribution between the instances and classes changes over time which causes concept drift problem. The concept drift is unique to data streams and makes the task of keeping classifier/model relevant difficult. As the concept changes, model performance may decrease, requiring a change or update in the training data. The concept drifted stream gives a new challenge to researcher that is the classes to be learned are not equally spread in the training data i.e. training data is imbalanced hence an another problem of class imbalance arises due to concept drift in data streams. This paper presents an algorithm ENSDS which is Ensemble based algorithm to handle Non-stationary Data Stream and we are comparing ours proposed algorithm with one of the existing Learn
⁺⁺
.NSE algorithm to show the performance improvement.