Forecasting with Exponential Smoothing

By Rob J Hyndman, Anne B. Koehler, J. Keith Ord, Ralph D. Snyder. 2008

1,439 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Time-Series & Forecasting

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Management Science and Operations Research

In Time-Series & Forecasting

Citations in the corpus

DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks (2017, methodology): “[11] provide a unifying review of these and related techniques.”
Forecasting Across Time Series Databases using Long Short-Term Memory Networks on Groups of Similar Series (2017, background): “…was able to win the CIF2016 forecasting competition for monthly time series (Štěpnička and Burda, 2016), outperforming state-of-the-art univariate algorithms such as ETS (Hyndman et al., 2008), BaggedETS (Bergmeir et al., 2016), Theta (Hyndman and Billah, 2003), and ARIMA (Box et al., 2015).”
Effective Bayesian Modeling of Groups of Related Count Time Series (2014, background): “Most classical time series forecasting models such as exponential smoothing (Hyndman et al., 2008) and ARIMA models (Box et al., 2008) assume that observations are realvalued and can take on both positive and negative values.”
Deep Factors for Forecasting (2019, methodology): “The second local model, DF-LDS, is a part of a special and robust family of SSMs, Innovation State-Space Models (ISSMs) (Hyndman et al., 2008; Seeger et al., 2016).”
Think Globally, Act Locally: A Deep Neural Network Approach to High-Dimensional Time Series Forecasting (2019, methodology): “For a comprehensive treatment of traditional methods, we refer the readers to [13, 19, 4, 18, 12] and the references there in.”
The Tourism Forecasting Competition (2011, methodology): “We then compute forecast intervals from the selected models using analytic formulae (Hyndman et al., 2008, Chapter 6), or by simulation if the analytic formulae are not available.”
Automatic Time Series Forecasting: The forecast Package for R (2008, methodology): “The linear exponential smoothing models are all special cases of ARIMA models—the equivalences are discussed in Hyndman et al. (2008a).”
Textual Explanations for Self-Driving Vehicles (2018, methodology): “At each time t, an acceleration, at, is measured by taking the derivative of speed measurements, and a change of course, ct, is computed by taking a difference between a current vehicle’s course and a smoothed value by using simple exponential smoothing method [10].”
Recurrent Neural Networks for Time Series Forecasting: Current Status and Future Directions (2019, methodology): “Time series forecasting has been traditionally a research topic in Statistics and Econometrics, from simple methods such as Seasonal Näıve and Simple Exponential Smoothing, to more complex ones such as ETS (Hyndman et al., 2008) and ARIMA (Box et al.”
Bagging exponential smoothing methods using STL decomposition and Box–Cox transformation (2016, background): “For more detailed descriptions, we refer to Hyndman et al. (2002), Hyndman et al. (2008), and Hyndman and Athanasopoulos (2013).”
Beyond point forecasting: evaluation of alternative prediction intervals for tourist arrivals (2011, methodology): “…time series models such as the AR model, the bias-corrected bootstrap for the AR model (Kim, 2004), the innovations state space models for exponential smoothing (as presented by Hyndman et al., 2008), Harvey's (1989) structural time series models, and the seasonal ARIMA model of Box et al. (1994).”
Statistical models underlying functions of 'smooth' package for R (2017, methodology): “The structure of state vector in our implementation differs from conventional structures (for example, in Hyndman et al. (2008)).”
Bayesian Intermittent Demand Forecasting for Large Inventories (2016, methodology): “Note that ISSMs are characterized by a single Gaussian innovation variable ε t per time step.”
Modeling and predicting behavioral dynamics on the web (2012, methodology): “We present methods for modeling behaviors over time that explain observed changes in the frequency of queries, clicked URLs, and clicked query-URL pairs.”
Predictability of monthly temperature and precipitation using automatic time series forecasting methods (2018, methodology): “The ‘forecast’ R package includes mostly methods based on exponential smoothing (Hyndman et al. 2008) or AutoRegressive Integrated Moving Average (ARIMA) (and related) stochastic processes.”
Forecasting with temporal hierarchies (2017, neutral): “The DGP we generate from is an ETS(A,Ad,A) model: µt = `t−1 + φbt−1 + st−m `t = `t−1 + φbt−1 + αεt bt = φbt−1 + βεt st = st−m + γεt ŷt+h|t = `t + φhbt + st−m+h+m where φh = φ+ · · ·+ φh and h+m = [ (h− 1) mod m ] + 1 (for more details see Hyndman et al., 2008).”
GluonTS: Probabilistic Time Series Models in Python (2019, methodology): “In contrast to ETS and ARIMA, which learn Φ per time series individually, neural generative models like (Rangapuram et al., 2018) further express Φ as a function of a neural network whose weights are shared among all time series and learned from the whole training data.”
An Experimental Review on Deep Learning Architectures for Time Series Forecasting (2020, methodology): “Before the rise of data mining techniques, the traditional methods used for TSF were mainly based on statistical models, such as exponential smoothing (ETS) 14 and Box-Jenkins methods like ARIMA.”