Frequent Subsequence Mining in Time Series Based on Symbolic Representation
Xiaoyun Chen · Jisuanji gongcheng · 2008
This paper proposes a new algorithm for mining frequent subsequence in time series based on symbolic representation.A dimensionality reduction technique called PAA linear segment representation is used.Under the Gaussian distribution,several breakpoints are set.The projected database is built to mine the frequent subsequence.The algorithm is simple and new,runs so fast,and reduces the cost of computing support counts of subsequences.