A Case-Based Model for Time Series Extrapolation
Qi Zheng · 2004
In this paper current time series similarity models are described in a unified way,based on which a general hierarchical model is proposed for reusing cases at several levels of abstraction,within a time series extrapolating con- text. For time series similarity assessment ,the paper shows how the case is represented,how cases are abstracted to higher lever and refined to lower lever ,and how similarity assessment is performed in a single hierarchical framework, followed by several discussion of several key issues in CBR circle. Finally ,an implementation on futures prediction us- ing this framework illustrates the idea we developed.