Furthest fitting cost: time series similarity from another angle

On time series similarity measurement, a new idea based on furthest fitting cost (FFC) is proposed, which cares the cost transforming one series to another one. This paper introduces amplitude scaling and y-axis shifting, to give a simple distance definition. Furthermore, a fast simple time warping method is proposed to deal with the time series of different lengths, to make FFC more applicable. With these models, a time series extrapolation solution is given, followed by the discussion of several issues. Experimental result shows that our approach is rather fast and reasonable.