Dot product based time series asynchronous periodic patterns mining algorithm
Cheng-Kui Gu, Xiaoli Dong · 2009
Mining periodic patterns in time-series databases is an interesting data-mining problem with wide application. Research on asynchronous periodic patterns is of great importance. The position list produce algorithm of each event is the essential prerequisite and foundation of the existing asynchronous periodic patterns mining algorithms. We propose a dot product based time series asynchronous periodic patterns detection algorithm. A binary representation based mapping scheme is designed, and a modified dot product algorithm is proposed to find all the positions of an event in the time series, which is a parallel calculation method replace the existing series calculation method, can notably decrease the times of the calculation. The experimental results show that our approach significantly increases the efficiency without loss of the accuracy.