An Efficient Indexing Structure for Group Models On Data Streams

Group learning is a common tool for data stream classification, mainly because of its inherent
advantages of handling huge volume of stream data and concept drifting. Have been mainly focused on
building accurate group models from stream data. a linear scan of a huge number of base classifiers in
the group during prediction incurs significant costs in response time, preventing group learning from
being practical for many real world time critical data stream applications, such as web traffic monitoring
spam detection, intrusion detection In these applications, data streams usually arrive at a speed of Giga
byte per Seconds, and it is necessary to classify each stream record in a timely manner we propose a
novel Ensemble tree indexing structure to organize all base classifiers in an grouped for fast prediction
On Ensemble trees treat group as spatial databases and employee an Random tree like height balanced
structure to minimize the expected prediction time of from linear to sub linear complexity. On the other
hand, Ensemble trees can be automatically updated by continuously integrating new classifiers and other
discarding outdated ones, well adapting to new trends and patterns underneath data streams.