We compare four methods of forecasting functional time series including fully functional regression, functional autoregression FAR(1) model, Hyndman & Shang principal component scores forecasting using one-dimensional time series method, and moving functional median. Our comparison methods involve simulation studies as well as analysis of empirical dataset concerning the Internet users behaviours for two Internet services in 2013. Our studies reveal that Hyndman & Shao predicting method outperforms other methods in the case of stationary functional time series without outliers, and the moving functional median induced by Frainman & Muniz depth for functional data outperforms other methods in the case of smooth departures from stationarity of the time series as well as in the case of functional time series containing outliers.
JavaScript is turned off in your web browser. Turn it on to take full advantage of this site, then refresh the page.