摘要：For modeling and proper and reliable parametric estimating of self-correlated data and time series, robust methods are used; because of the fact that existence of contaminated data and outliers, has an undesirable effect on estimation of parameters in these models. Since in most financial data past data is effective on recent data, these problems can be implemented by models of time series. In this paper, autoregressive models are considered as a model for the time series. A new robust method is presented based on filtered S optimization approach to estimate the parameters of autoregressive model. Resulted robust model can be used for robust prediction of the future values. Finally, as a numerical example, resulted profit of an intermediate product in 148 months is presented and suggested robust method is applied on it. Robust method, compared to classical methods, shows higher efficiency in predicting future values.
关键词：Time series;autoregressive model;outliers;robust estimation;financial data
