Data stream classification suffered from the problem of infinite length, concept drift,  concept-evolution and feature-evolution in data mining community. Usually data streams are infinite in length and it makes difficult to store and use all the historical data for training. Several research is eliminating the concept-evolution and feature-evolution concepts. Recently, many researchers have been focused on data streams as an important approach against huge database mining instead of mine the entire database. In this paper, an efficient approach is proposed for classification and novel class detection of data streams. The proposed method uses the outlier detection method to remove the unwanted data present on the data streams. Also, this approach uses the Nearest Neighbor algorithm and the Naive Bayes classifier concepts for novel class detection. The performance is evaluated with respect to error rates, final word count, speed and time. The result shows that the proposed method for classification and novel class detection provides better results than the existing techniques