Prediction of dynamical time series using kernel based regression and smooth splines (Q1657952)

The authors use the smooth splines for denoising and prediction of flows. They propose an algorithm for denoising and prediction in two steps. In the first step, the denoising is provided through smooth splines. In the second step, kernel based regression is used for the computation of the predictor \(f_1\). Beside this, the convergence of the computed predictor \(f_1\) to the exact predictor \(F\) is considered. At the end, three different methods for the computation of an approximated predictor are compared and the obtained numerical results are discussed in more detail.