A similarity-based approach for data stream classification

Highlights：

• A new approach for data streams classification using the instance based Learning techniques, is proposed.

• A new insertion/removal policy that adapts quickly to the data concept and maintains a small set of examples is proposed.

• The methodology is able to detect novel class, during the running phase, and remove unuseful ones.

• A large suit of data streams and statistical tests were used to evaluate the model performance.

• Results demonstrate that the proposed method is very competitive in terms of accuracy and time processing.