Financial Time Series Segmentation based on Specialized Binary Tree Representation

is one of the fundamental components in time series data mining. One of the uses of the time series segmentation is trend analysis-to segment the time series into primitive trends like uptrend and downtrend. In this paper, a time series segmentation method based on a specialized binary tree representation scheme is proposed; this representation scheme is customized for financial time series to cater for its unique behaviors. The proposed segmentation method is based on the concept of data point importance and the location of the cutting points is already encoded in the representation scheme. Therefore, no additional effect is needed to determine the cutting points. One may find it particularly attractive in applications like stock data analysis. The unique behavior of the proposed segmentation method is demostrated by applying to financial time series.