Enhancing DWT for Recent-Biased Dimension Reduction of Time Series Data
In many applications, old data in time series become less important as time elapses, which is a big challenge to traditional techniques for dimension reduction. To improve Discrete Wavelet Transform (DWT) for effective dimension reduction in this kind of applications, a new method, largest-latest-DWT, is designed by keeping the largest k coefficients out of the latest w coefficients at each level of DWT transform. Its efficiency and effectiveness is demonstrated by our experiments.
KeywordsTime series dimension reduction
Unable to display preview. Download preview PDF.
- 1.Bulut, A., Singh, A.K.: SWAT: hierarchical stream summarization in large networks. In: Proc. of ICDE 2003 (2003)Google Scholar
- 2.Fu, T.-C., Chung, F.-L., Ng, V., Luk, R.: Pattern discovery from stock time series using self-organizing maps. In: KDD 2001 Workshop on Temporal Data Mining (2001)Google Scholar
- 3.Fink, E., Pratt, K.B., Gandhi, H.S.: Indexing of time series by major minima and maxima. In: Proc. of IEEE Int. Conf. on Systems, Man and Cybernetics (2003)Google Scholar
- 5.Mörchen, F.: Time series feature extraction for data mining using DWT and DFT. Tech. Report No. 33, Dept. of Maths and CS, Philipps-University Marburg (2003)Google Scholar
- 6.Palpanas, T., Vlachos, M., Keogh, E., Gunopulos, D., Truppel, W.: Online amnesic approximation of streaming time series. In: Proc. of ICDE 2004 (2004)Google Scholar
- 7.Perng, C.-S., Wang, H., Zhang, S.R., Parker, D.S.: Landmarks: a new model for similarity-based pattern querying in time series databases. In: Proc. of ICDE 2000 (2000)Google Scholar