Robust Forecasting of Non-Stationary Time Series

C. Croux, R. Fried, I. Gijbels, K. Mahieu

Research output: Working paperDiscussion paperOther research output

407 Downloads (Pure)


This paper proposes a robust forecasting method for non-stationary time series. The time series is modelled using non-parametric heteroscedastic regression, and fitted by a localized MM-estimator, combining high robustness and large efficiency. The proposed method is shown to produce reliable forecasts in the presence of outliers, non-linearity, and heteroscedasticity. In the absence of outliers, the forecasts are only slightly less precise than those based on a localized Least Squares estimator. An additional advantage of the MM-estimator is that it provides a robust estimate of the local variability of the time series.
Original languageEnglish
Place of PublicationTilburg
Number of pages17
Publication statusPublished - 2010

Publication series

NameCentER Discussion Paper


  • Heteroscedasticity
  • Non-parametric regression
  • Prediction
  • Outliers
  • Robustness

Cite this