A Presentation of Time Series Algorithm Based on K-Mean and Its Application in Clustering
Download as PDF
As the data volume of a time series database is much greater than that of an ordinary database, some general data mining tools cannot be applied to time series directly to generate satisfactory result. Therefore, a piecewise linearization representation algorithm is proposed for time series in this paper, which can greatly improve the computation speed of similarity measurement. Based on the piecewise linearization representation, a similarity calculation method is proposed, which is insensitive to various deformation of time series. The k-mean clustering algorithm is applied to time series represented by piecewise linearization for a desired result.
Time series, Algorithm, Similarity, K-mean, Clustering