An Algorithm for Predictive Queries over Data Stream Based on Energy and Frequent Pattern
基于能量和频繁模式的数据流预测查询算法

A new predict model was contrived,which involves local stream energy prediction,the energy distribution pattern mining,the predictive series reconstruction and measurement method of stream energy.A new method was designed to forecast stream by energy regression and wavelets decomposing based on frequent pattern, and extended to multi-streams with strong coincidence.The concept of the nearest maximum frequent pattern was proposed to decompose local stream energy.The validity of new algorithm was demonstrated by extensive experiments on real data.