Research Information System
Thesis Type: Postgraduate
Institution Of The Thesis: Middle East Technical University, Faculty of Economic and Administrative Sciences, Department of Economics, Turkey
Approval Date: 2020
Student: Kadir Özen
Supervisor: DİLEM YILDIRIM KASAP
Abstract:Fundamental dynamics behind electricity prices are multi-dimensional and elaborate. A popular approach to forecasting electricity price is to utilize large number of predictors. In this study, using the day-ahead electricity price data from commonly studied markets of five major series and GEFCom2014 data, a variant of shrinkage method, Bootstrap Aggregation (bagging) is proposed to incorporate information from available predictors. Bagging manifests itself as a computationally simpler alternative to commonly used Least Absolute Shrinkage and Selection Operator (lasso) in multivariate EPF context and even shows superior forecasting ability in some markets. Moreover, considering the significant dependence of intra-day electricity prices, we also propose factor augmentation to exploit this dependence. The inclusion of latent factors, selected via Bayesian Information Criterion, improves ability to forecast in multivariate modeling framework and in some cases even outperform sophisticated shrinkage methods as measured by the Diebold-Mariano test.