《Segmentation of Time Series by the Clustering and Genetic Algorithms》.pdf

Segmentation of Time Series by the Clustering and Genetic Algorithms † † † ‡ Vincent S. Tseng , Chun-Hao Chen , Chien-Hsiang Chen and Tzung-Pei Hong †Department of Information Engineering, National Cheng-Kung University ‡ Department of Electrical Engineering, National University of Kaohsiung {vincent, chchen, chienhsiang}@.tw, tphong@.tw Abstract The segmentation problem is usually converted into an optimization problem. Recently, the genetic algorithm has This paper proposes a time series segmentation been proven to be very suitable for solving optimization approach by combining the clustering technique, the problems. Chung et al. thus proposed a GA-based discrete wavelet transformation and the genetic algorithm approach to segment time series [3]. They allowed users to automatically find segments and patterns from a time to specify a set of desired patterns and search for them series. The genetic algorithm is used to find the during the segmentation processing steps. They used the segmentation points for deriving appropriate patterns. In real-number schema to encode each chromosome, which the fitness evaluation, the proposed algorithm first divides represented a possible segmentation result. T