An Efficient Method for Discovering Motifs in Large Time Series

摘要

Time series motif is a previously unknown pattern appearing frequently in a time series. An efficient motif discovery algorithm for time series would be useful as a tool for summarizing massive time series databases as well as many other advanced time series data mining tasks. In this paper, we propose a new efficient algorithm, called EP-BIRCH, for finding motifs in large time series datasets. This algorithm is more efficient than MK algorithm and stable to the changes of input parameters and these parameters are easy to be determined through experiments. The instances of a discovered motif may be of different lengths and user does not have to predefine the length of the motif.