[关键词]
[摘要]
提出一种有效的时间序列流伪周期检测方法MPD(memory-constrain period detection).它采用Haar小波技术构建时间序列流大纲,利用部分片段估计周期方法提高检测效率,采用基于三次插值的周期估计方法检测任意长度的周期.通过对MPD误差的理论分析和实验分析,验证了MPD的时间和空间复杂度以及检测误差的有效性.
[Key word]
[Abstract]
A period detection method called MPD(memory-constrain period detection) is proposed naively on a time series stream, where the Haar-wavelet synopsis of series stream is adopted, and an estimated period based on partial fragments is proposed to improve the detection efficiency, and the cubic spline is used to detect period of arbitrary length. The time and space complexity error bound of MPD are validated through theoretical and experimental analysis.
[中图分类号]
[基金项目]
Supported by the National Natural Science Foundation of China under Grant Nos.60703068, 60873068 (国家自然科学基金)
