An Efficient Indexing Structure for Ensemble Classification of Data Streams Using Forest-Tree Mechanism

Computer Engineering

Ensemble learning is used for data stream classification, as it facing problem to large size of stream data and concept drifting. Direct output of an extensive number of base classifiers in the troupe amid expectation keeping group gaining from being viable for some true time critical data stream applications, e.g. Web traffic. In this data streams usually come at a speed of GBPS, and it is important to order every stream record in a timely manner. That’s why we propose a novel E-tree indexing structure to sort out all bases in an ensemble for fast prediction and using random forest classifier-trees regard groups as spatial databases and utilize an R-tree to less the expected prediction time from direct to sub-straight multifaceted nature. E-trees can be changed by continuously integrating new classifiers and discarding outdated ones, well adjusting to new patterns. Our system works on web traffic stream monitoring to classify web pages and extend work to classify traffic stream data.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.