Accumulated prediction errors, information criteria and optimal forecasting for autoregressive time series

DOI10.1214/009053606000001550MaRDI QIDQ2642748zbMATH OpenOpenAlexFDO

Authors Ching-Kang Ing

Publication date 4 September 2007

Published in The Annals of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/0708.2373

zbMATH Keywords

accumulated prediction errorsasymptotic efficiencyasymptotic equivalenceinformation criterionoptimal forecastingorder selection

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Time series, auto-correlation, regression, etc. in statistics (GARCH) (62M10) Inference from stochastic processes and prediction (62M20)

Abstract: The predictive capability of a modification of Rissanen's accumulated prediction error (APE) criterion, APE${}_{delt{a}_n}$, is investigated in infinite-order autoregressive (AR($infty$)) models. Instead of accumulating squares of sequential prediction errors from the beginning, APE${}_{delt{a}_n}$ is obtained by summing these squared errors from stage $ndelt{a}_n$, where $n$ is the sample size and $1/nleqdelt{a}_{n}leq1-(1/n)$ may depend on $n$. Under certain regularity conditions, an asymptotic expression is derived for the mean-squared prediction error (MSPE) of an AR predictor with order determined by APE${}_{delt{a}_n}$. This expression shows that the prediction performance of APE${}_{delt{a}_n}$ can vary dramatically depending on the choice of $delt{a}_n$. Another interesting finding is that when $delt{a}_n$ approaches 1 at a certain rate, APE${}_{delt{a}_n}$ can achieve asymptotic efficiency in most practical situations. An asymptotic equivalence between APE${}_{delt{a}_n}$ and an information criterion with a suitable penalty term is also established from the MSPE point of view. This offers new perspectives for understanding the information and prediction-based model selection criteria. Finally, we provide the first asymptotic efficiency result for the case when the underlying AR($infty$) model is allowed to degenerate to a finite autoregression.

Recommendations

• Accumulative prediction error and the selection of time series models
• A derivation of the information criteria for selecting autoregressive models
• Order selection for same-realization predictions in autoregressive processes
• Asymptotically efficient order selection of a nonzero mean AR process for h-step prediction.
• scientific article; zbMATH DE number 1034039

Cites work

• A new look at the statistical model identification
• A note on mean-squared prediction errors of the least squares predictors in random walk models
• Adaptive prediction by least squares predictors in stochastic regression models with applications to time series
• Asymptotic Inference about Predictive Ability
• Asymptotically efficient selection of the order by the criterion autoregressive transfer function
• Asymptotically efficient selection of the order of the model for estimating parameters of a linear process
• Estimating the dimension of a model
• scientific article; zbMATH DE number 3928147 (Why is no real title available?)
• scientific article; zbMATH DE number 4106081 (Why is no real title available?)
• scientific article; zbMATH DE number 3635352 (Why is no real title available?)
• scientific article; zbMATH DE number 1034040 (Why is no real title available?)
• scientific article; zbMATH DE number 1034045 (Why is no real title available?)
• scientific article; zbMATH DE number 777876 (Why is no real title available?)
• scientific article; zbMATH DE number 2199140 (Why is no real title available?)
• Least squares estimates in stochastic regression models with applications to identification and control of dynamic systems
• Model selection and prediction: Normal regression
• Nonasymptotic bounds for autoregressive time series modeling.
• On predictive least squares principles
• On same-realization prediction in an infinite-order autoregressive process.
• On the selection of forecasting models
• Order selection for AR models by predictive least squares
• Order selection for same-realization predictions in autoregressive processes
• Properties of Predictors in Misspecified Autoregressive Time Series Models
• Regression and time series model selection in small samples
• Robust out-of-sample inference
• Selection of the order of an autoregressive model by Akaike's information criterion
• Strong consistency of the PLS criterion for order determination of autoregressive processes
• THE ESTIMATION OF THE ORDER OF AN AUTOREGRESSION USING RECURSIVE RESIDUALS AND CROSS-VALIDATION
• VAR forecasting under misspecification

Cited in

(27)

• Model averaging prediction for time series models with a diverging number of parameters
• Consistent model selection criteria and goodness-of-fit test for common time series models
• Model selection for high-dimensional linear regression with dependent observations
• Order selection for possibly infinite-order non-stationary time series
• Model selection: a Lagrange optimization approach
• Accumulative prediction error and the selection of time series models
• Least-squares forecast averaging
• Banded regularization of autocovariance matrices in application to parameter estimation and forecasting of time series
• Estimating prediction error: cross-validation vs. accumulated prediction error
• Prediction errors in nonstationary autoregressions of infinite order
• Simultaneous confidence bands for sequential autoregressive fitting
• Model selection for weakly dependent time series forecasting
• Asymptotic theory of generalized information criterion for geostatistical regression model selection
• Cross-validation for selecting a model selection procedure
• Predictor selection for positive autoregressive processes
• Negative Moment Bounds for Stochastic Regression Models with Deterministic Trends and Their Applications to Prediction Problems
• Forecasting time series of economic processes by model averaging across data frames of various lengths
• Multistep forecast selection for panel data
• On asymptotic risk of selecting models for possibly nonstationary time-series
• Selection of linear mixed‐effects models for clustered data
• Parametric or nonparametric? A parametricness index for model selection
• Information criteria for model selection
• On model selection from a finite family of possibly misspecified time series models
• Model averaging prediction for possibly nonstationary autoregressions
• Model selection for integrated autoregressive processes of infinite order
• Toward optimal model averaging in regression models with time series errors
• Finite sample FPE and AIC criteria for autoregressive model order selection using same-realization predictions

This page was built for publication: Accumulated prediction errors, information criteria and optimal forecasting for autoregressive time series

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2642748)