Title: Modeling and forecasting nonlinear seasonality
Authors: Harry Haupt - University of Passau (Germany) [presenting]
Joachim Schnurbus - University of Passau (Germany)
Abstract: A nonparametric approach to modeling and forecasting multiple time series regressions with nonlinear seasonality is presented. A new kernel function allows us to flexibly take into account the specific periodic structure of seasonal effects. Smoothing parameters are estimated by a novel approach tailored to improve the kernel regression forecasting performance. The forecasting performance is discussed in an extensive Monte Carlo analysis and real data applications.