Classification of data streams with skewed distribution

One of fundamental problem in the task of mining streaming data is the concept drift over time. Such data Streams may also exhibit high and varying degrees of class imbalance, which can further complicate the task. In scenarios like these, class imbalance is particularly difficult to overcome and has not been as thoroughly studied. Most of the studies on classification of data streams assume relatively balanced and stable data streams but cannot handle well rather skewed streams which are typical in many data stream applications. Class imbalance in such skewed data streams can be seen in many real world applications. In such scenarios learning from skewed data streams results in classifier biased towards the majority class which results in misclassification of minority class examples, since in these scenarios minority class examples are too less than the majority class. The losses associated with misclassification of minority classes can be higher in some applications. In this paper we present our preliminary work to deal with classification of the data streams with skewed distribution of classes.