A Chaos-Based Predictive Algorithm for Continuous Aggregate Queries over Data Streams

It is very important to predict the future value or trend of continuous aggregate queries over data streams. But so far, many research works over data streams, especially from database field, have been focusing mainly on the approximate query processing. Although there are a few literatures to discuss the prediction issues of continuous aggregate queries over data streams, none of them allow for the effect that inherent characteristics of stream data itself imposed on prediction. The chaotic property shows some inherent principle, which would influence the stream´s future behaviors. Based on this, a novel chaos-based online predictive algorithm for continuous aggregate queries is proposed in this paper. This algorithm borrows local approximation prediction of chaotic time series to forecast the future values of stream data. The extensive experimental results show that the proposed algorithm has higher performance and provides better prediction of aggregate values over data streams.