An empirical comparison of alternative schemes for combining electricity spot price forecasts

Computing electricity spot price prediction intervals using quantile regression and forecast averaging

“…This fact is a good motivation for considering combining electricity spot price forecasts. Surprisingly, this approach has not been undertaken in the literature until very recently, see Bordignon et al (2013), Nowotarski et al (2014) and Raviv et al (2013). All three cited papers yield similar conclusions-they support the benefits of combining forecasts for deriving more accurate and more robust point forecasts of electricity spot prices.…”

“…Thus, the set of individual techniques considered here includes the 12 models analyzed by Weron and Misiorek (2008) and then used in the context of averaging point forecasts by Nowotarski et al (2014): autoregressive models (AR, ARX-the latter with temperature as the eXogenous variable), spike preprocessed autoregressive models (p-AR, p-ARX; where the model structure was estimated after replacing price spikes with less extreme observations), threshold autoregressive models (TAR, TARX), meanreverting jump diffusions (MRJD, MRJDX) and two classes of semiparametric autoregressive models (IHMAR, IHMARX, SNAR, SNARX; introduced by Weron and Misiorek 2008). In this study, we also use two of those individual models-ARX and SNARX-as benchmarks for comparison of the prediction intervals.…”

“…1. The prices and temperatures were preprocessed in the standard way for 'missing'/'doubled' values resulting from changes to/from the daylight saving time: the 'missing' values were interpolated from the two neighboring observations while the 'doubled' values were averaged to yield one observation for the doubled hour (as in Nowotarski et al 2014;Weron and Misiorek 2008). The logarithms of hourly temperatures T t (in Fahrenheit) were used as the exogenous variable in the time series models for the log-prices (no additional transformations were required since the temperatures were above 1 • F in the studied period).…”

Uncertainty analysis using Bayesian Model Averaging: a case study of input variables to energy models and inference to associated uncertainties of energy scenarios

“…The proposed quantification of uncertainty departs from a probabilistic assessment in Bayesian terms. The Bayesian Model Averaging (BMA) method is a well-established concept which already has been applied in energy economics [2,3]. By intention, the method presented is not novel and relies on accepted concepts and theories.…”

“…epistemic uncertainty. 3 Although assessment methods for epistemic uncertainty have been proposed in other contexts [34], these methods do not seem to be applicable in energy modelling. By epistemic uncertainties in the context of energy models I mean a form of under-determinism or "notknowability" of decisive influences on the energy system.…”