基于AR_TSM的时间序列motif关联规则挖掘方法研究

Research on time series motif association rule mining method based on AR_TSM

挖掘时间序列motif间潜在的关联规则可以在预测未来趋势方面发挥重要作用,时间序列motif即时间序列中先前未知的重复出现的模式。针对符号化时间序列提取motif导致信息丢失的问题,提出基于剪枝技术的motif提取算法PM_Motif,实现了保留原始信息的motif的精准快速提取;针对分割motif来发现其内部关联规则导致的规则不一致的问题,从motif间的关联规则入手,给出了基于AR_TSM方法的时间序列motif关联规则挖掘算法,从根本上避免了因motif分割引起的不确定性,保证了规则的一致性;最后,引入了关联规则评价参数RM,在多数据集上证明了关联规则的预测性能。

Mining potential association rules between time series motif can play an important role in predicting future trends.Time series motif is a previously unknown recurring pattern in time series.Aimed at the problem of information loss caused by symbolic time series when extract the motif,this paper developed a motif extraction algorithm PM_Motif based on pruning technology.It could extract motif accurately and quickly.Aimed at the problem of inconsistent rules caused by splitting motif to find its internal association rules,this paper developed an association rule mining algorithm between time series motif based on AR_TSM method.It could fundamentally avoid the uncertainty caused by motif segmentation and ensure the consistency of the rules.Finally,it introduced the association rule evaluation parameter RM,and could prove the prediction performance of the association rule mining algorithm on multiple datasets.