A New Clustering Algorithm for Time Series Analysis

Abstract

Conventional model-based clustering algorithms for time series data are limited in improving the clustering performance and also their computation complexity is high. In order to tackle this problem, a new model-based clustering algorithm with a certainty factor is proposed to evaluate the certainty degree of time series data being in a cluster. The new algorithm can be used to show a reasonable result for time series data clustering and reduce the computation complexity greatly. Performance of the algorithm is verified by the experiments on both synthetic data and real data.