Improving Recurrent Forecasting in Singular Spectrum Analysis usingKalman Filter Algorithm

One of the most practical nonparametric methods in the analysis of time series observations is the singular spectrum analysis (SSA) method. This method has been developed and applied to many practical problems across different fields and continuous efforts have been made to improve this method, especially in forecasting. In this paper the state space model and Kalman filter algorithms are used for noise elimination and time series smoothing. Finally, we compare these forecasting methods’ abilities using the root mean squared error criterion (RMSE) for simulation studies and real data.