Forecasting using random subspace methods

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Random subspace methods are a new approach to obtain accurate forecasts in high-dimensional regression settings. Forecasts are constructed by averaging over forecasts from many submodels generated by random selection or random Gaussian weighting of predictors. This paper derives upper bounds on the asymptotic mean squared forecast error of these strategies, which show that the methods are particularly suitable for macroeconomic forecasting. An empirical application to the FRED-MD data confirms the theoretical findings, and shows random subspace methods to outperform competing methods on key macroeconomic indicators.

Original languageEnglish
Pages (from-to)391-406
Number of pages16
JournalJournal of Econometrics
Volume209
Issue number2
DOIs
Publication statusPublished - Apr 2019

Keywords

  • Dimension reduction
  • Forecasting
  • Random subspace

Cite this

Boot, Tom ; Nibbering, Didier. / Forecasting using random subspace methods. In: Journal of Econometrics. 2019 ; Vol. 209, No. 2. pp. 391-406.
@article{959c82f6a9b1435b8ccff48510e5aa6f,
title = "Forecasting using random subspace methods",
abstract = "Random subspace methods are a new approach to obtain accurate forecasts in high-dimensional regression settings. Forecasts are constructed by averaging over forecasts from many submodels generated by random selection or random Gaussian weighting of predictors. This paper derives upper bounds on the asymptotic mean squared forecast error of these strategies, which show that the methods are particularly suitable for macroeconomic forecasting. An empirical application to the FRED-MD data confirms the theoretical findings, and shows random subspace methods to outperform competing methods on key macroeconomic indicators.",
keywords = "Dimension reduction, Forecasting, Random subspace",
author = "Tom Boot and Didier Nibbering",
year = "2019",
month = "4",
doi = "10.1016/j.jeconom.2019.01.009",
language = "English",
volume = "209",
pages = "391--406",
journal = "Journal of Econometrics",
issn = "0304-4076",
publisher = "Elsevier",
number = "2",

}

Forecasting using random subspace methods. / Boot, Tom; Nibbering, Didier.

In: Journal of Econometrics, Vol. 209, No. 2, 04.2019, p. 391-406.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Forecasting using random subspace methods

AU - Boot, Tom

AU - Nibbering, Didier

PY - 2019/4

Y1 - 2019/4

N2 - Random subspace methods are a new approach to obtain accurate forecasts in high-dimensional regression settings. Forecasts are constructed by averaging over forecasts from many submodels generated by random selection or random Gaussian weighting of predictors. This paper derives upper bounds on the asymptotic mean squared forecast error of these strategies, which show that the methods are particularly suitable for macroeconomic forecasting. An empirical application to the FRED-MD data confirms the theoretical findings, and shows random subspace methods to outperform competing methods on key macroeconomic indicators.

AB - Random subspace methods are a new approach to obtain accurate forecasts in high-dimensional regression settings. Forecasts are constructed by averaging over forecasts from many submodels generated by random selection or random Gaussian weighting of predictors. This paper derives upper bounds on the asymptotic mean squared forecast error of these strategies, which show that the methods are particularly suitable for macroeconomic forecasting. An empirical application to the FRED-MD data confirms the theoretical findings, and shows random subspace methods to outperform competing methods on key macroeconomic indicators.

KW - Dimension reduction

KW - Forecasting

KW - Random subspace

UR - http://www.scopus.com/inward/record.url?scp=85061570888&partnerID=8YFLogxK

U2 - 10.1016/j.jeconom.2019.01.009

DO - 10.1016/j.jeconom.2019.01.009

M3 - Article

VL - 209

SP - 391

EP - 406

JO - Journal of Econometrics

JF - Journal of Econometrics

SN - 0304-4076

IS - 2

ER -