Forecasting seasonals and trends by exponentially weighted moving averages

By Charles C. Holt. 2004

1,855 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Statistics and Probability

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 Statistics and Probability

Citations in the corpus

N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019, methodology): “The statistical modeling approaches based on exponential smoothing and its different flavors are well established and are often considered a default choice in the industry [10, 11, 31].”
Multi-Horizon Time Series Forecasting with Temporal Attention Learning (2019, methodology): “Classical time series forecasting approaches includeHolt-Winters method [15, 30] and ARIMA [5].”
A Double-Layer Neural Network Framework for High-Frequency Forecasting (2017, background): “This group of indicators describes the market momentum [Holt 2004].”
Damped trend exponential smoothing: A modelling viewpoint (2010, positive): “In a series of three papers (Gardner & McKenzie, 1985, 1988, 1989), we developed new versions of the Holt–Winters (Holt, 2004; Winters, 1960) methods of exponential smoothing that damp the trend as the forecast horizon increases.”
25 Years of IIF Time Series Forecasting: A Selective Review (2005, methodology): “These methods originated in the 1950s and 1960s with the work of Brown (1959, 1963), Holt (1957, 2004) and Winters (1960).”
Advances in Forecasting Under Instability (2012, neutral): “The initial value for the recursion can be the initial observation.27 Holt (1957) and Winters (1960) generalized the approach to include a local linear trend.”
Predicting Taxi–Passenger Demand Using Streaming Data (2013, methodology): “The weight set ω is calculated using a well-known time-series approach to these type of problems, i.e., the exponential smoothing [33].”
Correcting Forecasts with Multifactor Neural Attention (2016, methodology): “Holt-Winters or ARIMA models require a minimum of two observation periods worth of data in order to provide an initial fit (in our case, two years worth of data).”
Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift (2022, methodology): “As an example of the statistical models, exponential smoothing forecasting (Holt, 2004; Winters, 1960) is a well-established benchmark for predicting future values.”
TSMixer: An all-MLP Architecture for Time Series Forecasting (2023, background): “A more general assumption is that the time series can be decomposed into a periodic sequence and a sequence with smooth trend (Holt, 2004; Zhang & Qi, 2005; Wu et al., 2021; Zhou et al., 2022b).”
Forecasting China’s GDP at the pixel level using nighttime lights time series and population images (2017, methodology): “In light of the use of a recursive calculation algorithm, exponential smoothing can smooth and forecast time series data without need of establishing parametric models (Gelper, Fried, and Croux 2010; Holt 2004).”
Dynamic Bradley–Terry modelling of sports tournaments (2013, background): “Dixon and Coles (1997) proposed an ad hoc adjustment of the Poisson distribution introducing a dependence parameter that modifies the probabilities of the results 0–0, 0–1, 1–0 and 1–1.”
EXPONENTIAL SMOOTHING: THE STATE OF THE ART, PART II (2006, positive): ““generalized Holt.” As Taylor (2003a) observed, generalized Holt is a clumsy way to model a”
A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem (2017, background): “Fortunately, it is believed that the correlation between two market price events decades exponentially with the temporal distance between them (Holt, 2004; Charles et al., 2006).”
The Insecurity of Home Digital Voice Assistants - Amazon Alexa as a Case Study (2017, methodology): “1) CSI Processing Phase: This phase consists of three modules: principal component analysis (PCA) [13], median and exponential moving average (EMA) ﬁlter [14], [15], and Butterworth low-pass ﬁlter [16].”