Authors: G.N.V.G. Sirisha; M. Shashi
Addresses: Department of CSE, S.R.K.R. Engineering College, Bhimavaram, India ' Department of CS and SE, A.U. College of Engineering, Andhra University, Visakhapatnam, India
Abstract: Periodic patterns occur in a wide variety of datasets like time series, temporal, spatiotemporal and biological datasets. Asynchronous periodic pattern mining discovers the patterns that appear significantly with strict periodicity in one or more subsequences called valid segments with tolerable disturbance between the valid segments. The state-of-the-art asynchronous periodic pattern mining algorithms proposed in the literature mine many redundant asynchronous periodic patterns. This paper proposes an algorithm to mine non-redundant asynchronous periodic patterns. It is a variation of a four-phase algorithm named SMCA. The proposed algorithm mines precise and concise set of non-redundant patterns whose number is in orders of magnitude smaller than that of previous methods. The proposed algorithm is explained using a hypothetical dataset and its performance is evaluated using three real data sets which are stock market data from Bombay Stock Exchange, Weather data from Cambridge University and Migratory Zebras' data from MoveBank.
Keywords: asynchronous periodic patterns; non-redundant patterns; hash-based technique; multivariate time series; event set sequence; generic patterns; weather data; migratory zebras data; stock market data; stock markets; data mining; pattern mining.
International Journal of Business Intelligence and Data Mining, 2016 Vol.11 No.3, pp.205 - 228
Received: 07 Jul 2016
Accepted: 05 Oct 2016
Published online: 29 Jan 2017 *