摘要: 在k-近邻局部异常检测算法的基础上,结合时间序列的分割方法,提出一种高效率的时间序列增量异常模式检测算法。将时间序列按序列重要点进行数据分割,利用局部异常检测方法检测出时间序列的异常模式。当插入一些新数据时,邻近分割模式发生变化,增量异常检测算法更新相应的最近邻模式。通过该算法可以高效率地发现时间序列的异常模式。
关键词:
时间序列,
增量异常模式,
局部异常因子
Abstract: An efficient incremental outlier pattern detection algorithm is proposed based on the foundation of k-nearest local outlier pattern detection algorithm and segmentation. Time series data are segmented using series important point in this algorithm. Outlier pattern is detected by local outlier detection technique. When some new data points are inserted, neighbors segment pattern is changed and incremental outlier detection algorithm only refreshes limited number of their nearest neighbors pattern. The outlier pattern can be efficiently detected from time series by this algorithm.
Key words:
time series,
incremental outlier pattern,
local outlier factor
中图分类号:
周大镯;刘 雷. 时间序列增量异常模式检测算法[J]. 计算机工程, 2009, 35(16): 45-47.
ZHOU Da-zhuo; LIU Lei. Time Series Incremental Outlier Pattern Detection Algorithm[J]. Computer Engineering, 2009, 35(16): 45-47.