Abstract: Focused on the present status that the algorithm of measuring time series distance is many but the appropriate algorithm of measuring nonuniform time series is not available，the paper puts forward an algorithm of measuring nonuniform time series distances based on the ideal of searching minimal distance of the side collection which is constructed by the global points of series.Both the algorithm realization of creating event series and that of measuring nonuniform time series distances are given out.Finally，the algorithms are tested by the data from UCI KDD for time series testing.The test result proves that the algorithm of measuring nonuniform time series distances can effectively reflect the similarities of nonuniform time series.

Key words: nonuniform time series, distance, clustering, data mining

摘要： 针对时间序列距离度量的算法很多，但没有适用于不规则时序距离度量算法的现状，基于寻求全局序列点构成的边集之间的距离路径最小的思想，提出一种不规则时序距离度量的算法，并给出了事件序列生成算法和不规则时序距离度量算法的实现，最后利用UCI KDD的时间序列测试数据对算法进行了测试。测试结果证明了该不规则时序距离算法能够有效度量不规则时序的相似性。

关键词: 不规则时间序列, 距离, 聚类, 数据挖掘