University of Melbourne, Australia - Department of Computer Science and Software Engineering
We propose a new method for clustering time series.
A univariate time series can be represented by a fixed-length vector whose components are statistical features of the time series, capturing the global structure. These descriptive vectors, then being clustered using a standard fast clustering algorithms. A further search mechanism is used to find the best selection from the features for some specific problem domain or data set. We demonstrate the effectiveness and simplicity of our proposed method by clustering some benchmark datasets with empirical results.
Extension based on this work from univariate to multivariate time series clustering on real-world data set will also be discussed.