Research  |  Consulting  |  Seminars/Courses  |  Advertising  |  About  |  Contact  |  News
Research > Macro & Forecasting > [WP04/2014] Variable Selection for Large Unbalanced Datasets Using Non-Standard Optimisation of Information Criteria and Variable Reduction Methods 

[WP04/2014] Variable Selection for Large Unbalanced Datasets Using Non-Standard Optimisation of Information Criteria and Variable Reduction Methods

 


Download paper from our server
Download paper from
SSRN

Variable Selection for Large Unbalanced Datasets Using Non-Standard Optimisation of Information Criteria and Variable Reduction Methods

George Kapetanios, Massimilano Marcellino & Fotis Papailias

quantf research Working Paper Series: WP04/2014


Abstract. We consider forecasting key macroeconomic variables using many predictors extracted from the Eurostat PEEIs dataset. To avoid the curse of dimensionality, we rely on model selection and model reduction. For model selection we use heuristic optimisation of information criteria, including simulated annealing, genetic algorithms, MC3 and sequential testing. For model reduction we employ the methods of principal components, partial least squares and Bayesian shrinkage regression. The problem of unbalanced datasets is discussed and potential solutions are suggested. We provide adequate evidence that these methods could be useful in forecasting. Their predictive performance is evaluated in a pseudo out-of-sample exercise, comparing the results relative to a univariate AR(1) benchmark. Our findings are very encouraging for forecasting the growth rate of quarterly consumption and GDP, and monthly industrial production and inflation.

Keywords. Heuristic optimisation, Information criteria, Unbalanced datasets, Model Reduction, Forecasting, PEEI

SSRN logo belongs to Social Science Electronic Publishing, Inc.


  
F. Papailias - D. Thomakos, (c) 2014
 
 
ETF Global Trading
 
ETF Momentum Trading
ETF Pairs Trading
FOREX Trading
 
Macroeconomic & Long-Term Forecasting
Research & Consulting
 
  
Disclaimer: The contents of quantf research (c) website (http://www.quantf.com) are provided for research and information purposes only. Prices, returns, strategy recommendations and all statistical estimates in general shown in this webpage are indicative and the authors are not offering to buy or sell or soliciting offers to buy or sell any financial instrument. The views in this website are those of the authors alone and are subject to change at any time. The authors of this webpage do not accept any liability whatsoever for any direct or consequential loss arising from any use of the information provided. The information in this webpage is not intended to predict actual results, which may differ substantially from those presented.