A non-Distance Measure of similarity between Uncertain Time series
Over past decades,there has been more and more large scale sensors deployed in a wide range of application areas.Due to the innate imprecision of sensor observations,researchers devote to find an efficient way to manage querying,mining and storing such uncertain data,in particular,the uncertain time series data meet with much recognition.Traditional measures such as Euclidean distance are not quite effective for analyzing uncertain time series.Up until recently,some measures have been proposed to process uncertain time series; However,they have limitations,for example,some of them provide less intuition of comparison between two uncertain time series and do not easily support multiple error distributions.We propose a new model of uncertain time series and present a new measure-O2IM2,which employ the observations interval and central tendency to measure the similarity of uncertain time series effectively.An extensive experimental evaluation is constructed to analyze our approach and make a comparison with the prior work on performance.Based on the evaluation,we improve the match accuracy and validate the effectiveness for uncertain time series.
Uncertian time series Similarity measure Observations interval Central tendency
Wei Wang Guohua Liu
School of Information Science and Technology,DongHua University,Shanghai 201620,China School of Computer Science and Technology,DongHua University,Shanghai 201620,China
国内会议
济南
英文
1-12
2014-10-16（万方平台首次上网日期，不代表论文的发表时间）