Forecasting, Structural Time Series Models and the Kalman Filter.

By Denise R. Osborn, Andrew C. Harvey. 1991

How researchers see this paper

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

Citations in the corpus

Deep Factors for Forecasting (2019, methodology): “The core of these methods is formed by comparatively simple (often linear) models, which require manual feature engineering and model design by domain experts to achieve good performance (Harvey, 1990).”
The Tourism Forecasting Competition (2011, methodology): “There are a large number of potential ARIMA models—too many to allow us to estimate every possible combination of p, q, P and Q. Instead, we need a way to efficiently traverse the space of models in order to arrive at the model with the lowest AIC value. Hyndman and Khandakar (2008) proposed the fol”
Forecasting Time Series With Complex Seasonal Patterns Using Exponential Smoothing (2011, methodology): “(1970) and Harvey (1989). Some of the key advantages of the TBATS modeling 173 framework are: (i) it admits a larger effective parameter space with the possibility of better 174 forecasts (Hyndman et al.”
The Multi-State Latent Factor Intensity Model for Credit Rating Transitions (2008, unknown): “Another valid alternative is the self-excitation mechanism introduced by Hawkes (1971) and also considered for the LFI model by Bauwens and Hautsch (2003).”
Predicting the present with Bayesian structural time series (2014, methodology): “Our system uses a structural time series model (Harvey, 1989) to capture the trend, seasonal, and similar components of the target series.”
Carbon emissions and oil consumption in Saudi Arabia (2015, methodology): “[10] and Dilaver and Hunt [7,8] argue that the structural time series modeling (STSM) approach developed by Harvey [14,13] is the appropriate methodology to efficiently capture the effect of unobservable factors (trends in energy efficiency, technological progress and consumer preference).”
Are Transformers Effective for Time Series Forecasting? (2022, background): “As a long-standing problem with a wide range of applications, statistical approaches (e.g., autoregressive integrated moving average (ARIMA) [1], exponential smoothing [12], and structural models [14]) for time series forecasting have been used from the 1970s onward.”
Machine Learning for the Geosciences: Challenges and Opportunities (2017, methodology): “Some of the existing methods for time-series forecasting include exponential smoothing techniques [62], autoregressive integrated moving average (ARIMA) models [63], state-space models [64], and probabilistic models such as hidden Markov models and Kalman ﬁlters [65], [66].”
BRITS: Bidirectional Recurrent Imputation for Time Series (2018, methodology): “Speﬁcically, we randomly generate a time series with length T = 36 , using the state-space representation [15]: where x t is the t -th value in time series.”
No-Arbitrage Taylor Rules (2007, methodology): “We use a Kalman filter algorithm that includes time-varying exogenous variables in the state equation following Harvey (1989).”
Estimating Macroeconomic Models: A Likelihood Approach (2007, unknown): “(19) Equation (19) is nearly identical to the likelihood function implied by the Kalman filter (see equation 3.4.5 in Harvey, 1989) when applied to a linear model.”
A protocol for data exploration to avoid common statistical problems (2010, background): “2009a) or allowing regression parameters to change over time (Harvey 1989).”