摘要
为了有效计算同一时间序列中具有不同长度的子序列数据的相似度,深入研究了时间序列中有关相似度计算的内容,提出了不等长子时间序列的相似性度量方法。该方法在分段计算每一相对独立子序列斜率的基础之上,能够有效地获得每一段的变化趋势,进而可以不受时间序列长度的影响而完成相似度计算。该方法在不同的数据集上进行实验,都获得了良好的效果,验证了该方法具有快速、准确、不受时间序列长度影响的特点。
In order to effectively compute the similarity of the sub-time series with different length from one time series,we deeply research the content about similarity computation in time series,and propose a similarity measurement method for sub-time series with different length.Based on computing the piecewise slope of each relatively independent sub-time series,the method can get their variation tendency,and then complete similarity computing without the influence of time series length.The method achieves good result in the experiments on different data sets,and its characteristics of speediness,accuracy and not influenced by time series length are proved.
出处
《计算机时代》
2011年第5期17-20,共4页
Computer Era
