Modeling and Forecasting Monotone Curves by FDA
A new estimation method and forecasting of monotone sample curves is performed from observations in a finite set of time points without a previous transformation of the original data. Monotone spline cubic interpolation is proposed for the reconstruction of the sample curves. Then, the interpolation basis is adapted to apply FPCA and forecasting is done by means of principal components prediction.
Unable to display preview. Download preview PDF.
- 1.Aguilera, A. M., Oca˜na, F. A., Valderrama, M. J.: An approximated principal component prediction model for continuous-time stochastic processes. Appl. Stoch. Model D. A. 13, 61–72 (1997)Google Scholar
- 2.Aguilera, A. M., Oca˜na, F. A., Valderrama, M. J.: Forecasting time series by functional PCA. Discussion of several weighted approaches. Computation. Stat. 14, 443–467 (1999)Google Scholar
- 4.Fritsch, F. N., Carlson, R. E.:Monotone piecewise cubic interpolation. SIAM J. Numer. Anal. 17, 238–246 (1980)Google Scholar
- 8.Valderrama, J. M., Aguilera, A. M., Oca˜na, F. A.: Predicci´on din´amica mediante an´alisis de datos funcionales. Ed. Hesp´erides-La Muralla, Madrid 2000)Google Scholar