Discovering All-Chain Set in Streaming Time Series

Time series chains discovery is an increasingly popular research area in time series mining. Previous studies on this topic process fixed-length time series. In this work, we focus on the issue of all-chain set mining over the streaming time series, where the all-chain set is a very important kind of the time series chains. We propose a novel all-chain set mining algorithm about streaming time series (ASMSTS) to solve this problem. The main idea behind the ASMSTS is to obtain the mining results at current time-tick based on the ones at the last one. This makes the method more efficiency in time and space than the Naive. Our experiments illustrate that ASMSTS does indeed detect the all-chain set correctly and can offer dramatic improvements in speed and space cost over the Naive method.

[1]  Eamonn J. Keogh,et al.  Matrix Profile II: Exploiting a Novel Algorithm and GPUs to Break the One Hundred Million Barrier for Time Series Motifs and Joins , 2016, 2016 IEEE 16th International Conference on Data Mining (ICDM).

[2]  João Gama,et al.  A survey on concept drift adaptation , 2014, ACM Comput. Surv..

[3]  Jiawei Han,et al.  Mining periodic behaviors of object movements for animal and biological sustainability studies , 2011, Data Mining and Knowledge Discovery.

[4]  Piotr Indyk,et al.  Motif discovery in physiological datasets: A methodology for inferring predictive elements , 2010, TKDD.

[5]  Eamonn J. Keogh,et al.  Matrix Profile VII: Time Series Chains: A New Primitive for Time Series Data Mining (Best Student Paper Award) , 2017, 2017 IEEE International Conference on Data Mining (ICDM).

[6]  Eamonn J. Keogh,et al.  Mining motifs in massive time series databases , 2002, 2002 IEEE International Conference on Data Mining, 2002. Proceedings..

[7]  Tim Oates,et al.  Finding story chains in newswire articles , 2012, 2012 IEEE 13th International Conference on Information Reuse & Integration (IRI).

[8]  Eamonn J. Keogh,et al.  Probabilistic discovery of time series motifs , 2003, KDD '03.

[9]  Eamonn J. Keogh,et al.  Matrix Profile I: All Pairs Similarity Joins for Time Series: A Unifying View That Includes Motifs, Discords and Shapelets , 2016, 2016 IEEE 16th International Conference on Data Mining (ICDM).

[10]  Manish Marwah,et al.  Visual exploration of frequent patterns in multivariate time series , 2012, Inf. Vis..

[11]  Eamonn J. Keogh,et al.  Time Series Chains: A Novel Tool for Time Series Data Mining , 2018, IJCAI.

[12]  Michael Gertz,et al.  Time will Tell: Temporal Linking of News Stories , 2015, JCDL.

[13]  Eamonn J. Keogh,et al.  Introducing time series chains: a new primitive for time series data mining , 2018, Knowledge and Information Systems.

[14]  Eamonn J. Keogh,et al.  Discovery of Meaningful Rules in Time Series , 2015, KDD.

[15]  Eamonn J. Keogh,et al.  Rare Time Series Motif Discovery from Unbounded Streams , 2014, Proc. VLDB Endow..