EFFICIENT MINING OF LOCAL FREQUENT PERIODIC PATTERNS IN TIME SERIES DATABASE

Recently, periodic pattern mining from time series data has been studied extensively. Existing studies on periodic patterns mining mainly consider discovering full periodic patterns from an entire time series. However, partial periodic patterns are more useful in practice since only some of the time episodes may exhibit periodic patterns. This paper aims to discover the partial periodic pattern in locality of the time series data. The notion of character locality is introduced to divide the time series into variable-length segments. We propose a novel algorithm, called LEPMiner, to find the local frequent periodic patterns in time series data. Experimental results show that the proposed algorithm is effective and efficient to reveal interesting local frequent periodic patterns.