“…So far, SM has been applied in a wide variety of predictive fields, e.g. finance (Chiu, 2002;Li & Ho, 2009;Oh & Kim, 2007;Shin & Han, 1999), manufacturing (Lee & Ha, 2009), agriculture Shih, Huang, Chiu, Chiud, & Hu, 2009), software development (Yang & Hsu, 2009) and medicine (Ahn & Kim, 2008;Elter, Wendtland, & Wittenberg, 2007). Nevertheless, SM belongs to the category of point-to-point computation algorithm, thus it is helpless in non-stationary time series prediction where the distribution of time series is changing over time.…”