A new data stream classification algorithm

Abstract

In data mining area, data stream classification, detecting concept drifts and updating temporary models are challenging tasks. To deal with this, big sample buffer and complex updating process are always needed for most of the current algorithms. In this article, a digital hormone based classification algorithm was presented. With the given way, we do not need a big sample-buffer in the classification process and the classifier can be updated efficiently. Experiments have shown that the proposed algorithm has the ability to predict the class label accurately and to store temporary records with more smaller memory space.