B0537
Title: Noise reduction for functional time series
Authors: Bram Wouters - University of Amsterdam (Netherlands) [presenting]
Abstract: A novel method for noise reduction in the setting of curve time series with error contamination is proposed, based on extending the framework of functional principal component analysis (FPCA). The underlying, finite-dimensional dynamics of the functional time series are employed to separate the serially dependent dynamical part of the observed curves from the noise. Upon identifying the subspaces of the signal and idiosyncratic components, a projection of the observed curve time series along the noise subspace is constructed, resulting in an estimate of the underlying denoised curves. This projection is optimal in the sense that it minimizes the mean integrated squared error. By applying the method to simulated and real data, the denoising estimator is shown consistent and outperforms existing denoising techniques. Furthermore, it is shown that it can be used as a pre-processing step to improve forecasting.
