An approach based on TSA-tree for accurate time series classification

Abstract

In order to improve the performance of time series classification, we introduce a new approach of time series classification. The first step of the approach is to design a feature exaction model based on Trend and Surprise Abstraction tree (TSA-tree). The second step of the approach is to combine the exacted global feature and 1 nearest neighbor to classify time series. The proposed approach is compared with a number of known classifiers by experiments in artificial and real-world data sets. The experimental results show it can reduce the error rates of time series classification, so it is highly competitive with previous approaches.