zbMATH — the first resource for mathematics

Robust forecasting with exponential and Holt-Winters smoothing. (English) Zbl 1203.62164
Summary: Robust versions of the exponential and Holt-Winters [see P. R. Winters, Manage. Sci. 6, 324–342 (1960; Zbl 0995.90562)] smoothing method for forecasting are presented. They are suitable for forecasting univariate time series in the presence of outliers. The robust exponential and Holt-Winters smoothing methods are presented as recursive updating schemes that apply the standard technique to pre-cleaned data. Both the update equation and the selection of the smoothing parameters are robustified. A simulation study compares the robust and classical forecasts. The presented method is found to have good forecast performance for time series with and without outliers, as well as for fat-tailed time series and under model misspecification. The method is illustrated using real data incorporating trend and seasonal effects.

62M20 Inference from stochastic processes and prediction
62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH)
65C60 Computational problems in statistics (MSC2010)
Full Text: DOI
[1] Bowerman, Forecasting, Time Series, and Regression (2005)
[2] Cipra, Robust exponential smoothing, Journal of Forecasting 11 pp 57– (1992) · Zbl 0951.91508
[3] Cipra, Kalman filter with outliers and missing observations, Test 6 pp 379– (1997) · Zbl 0893.62094
[4] Davies, Robust signal extraction for on-line monitoring data, Journal of Statistical Planning and Inference 122 pp 65– (2004) · Zbl 1040.62099
[5] Fried, Robust filtering of time series with trends, Nonparametric Statistics 16 pp 313– (2004) · Zbl 1065.62162
[6] Gardner, Note: rule-based forecasting vs. damped-trend exponential smoothing, Management Science 45 pp 1169– (1999) · Zbl 1231.91380
[7] Gardner, Exponential smoothing: the state of the art. Part II, International Journal of Forecasting 22 pp 637– (2006)
[8] Holt C. 1959. Forecasting seasonals and trends by exponentially weighted moving averages. ONR Research Memorandum 52.
[9] Hyndman, Prediction intervals for exponential smoothing using two new classes of state space models, Journal of Forecasting 24 pp 17– (2005)
[10] Kirkendall, Monitoring for outliers and level shifts in kalman filter implementations of exponential smoothing, Journal of Forecasting 11 pp 543– (2006)
[11] Kotsialos, Long-term sales forecasting using Holt-Winters and neural network methods, Journal of Forecasting 24 pp 353– (2005)
[12] Kutner, Applied Linear Statistical Models (2005)
[13] Makridakis, Forecasting, Methods and Applications (1998)
[14] Maronna, Robust Statistics, Theory and Methods (2006)
[15] Martin, Robust-resistant spectrum estimation, Proceedings of the IEEE 70 pp 1097– (1982)
[16] Masreliez, Robust Bayesian estimation for the linear model and robustifying the kalman filter, IEEE Transactions on Automatic Control 22 pp 361– (1977) · Zbl 0354.93054
[17] Pearson, Mining Imperfect Data: Dealing with Contamination and Incomplete Records (2005) · Zbl 1079.62005
[18] Romera, On practical implementation of robust kalman filtering, Communications in Satistics 24 pp 461– (1995) · Zbl 0850.62688
[19] Ruckdeschel P. 2001. Ansatze zur Robustifizierung des Kalman-Filters. Dissertation, Bayreuther Mathematische Schriften, Bayreuth. · Zbl 0995.93001
[20] Siegel, Robust regression using repeated medians, Biometrika 68 pp 242– (1982) · Zbl 0483.62026
[21] Taylor, Smooth transition exponential smoothing, Journal of Forecasting 23 pp 385– (2004)
[22] Taylor, Forecasting daily supermarket sales using exponentially weighted quantile regression, European Journal of Operational Research 178 pp 154– (2007) · Zbl 1102.62103
[23] Winters, Forecasting sales by exponentially weighted moving averages, Management Science 6 pp 324– (1960) · Zbl 0995.90562
[24] Yohai, High breakdown estimates of regression by means of the minimization of an efficient scale, Journal of the American Statistical Association 83 pp 406– (1988) · Zbl 0648.62036
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.