Do. 28.08.2014, Kolloquium
Thema: Forecasting generalized quantiles of electricity demand: A functional data approach, Referent: Franziska Schulz (Humboldt-Universität Berlin), Zeit: 14:00 Uhr, Raum: ENC D-223.
Mathematisches Kolloquium
Donnerstag, den 28.08.2014, 14:00 Uhr, Raum ENC D-223
Thema: Forecasting generalized quantiles of electricity
demand: A functional data approach
Referent: Franziska Schulz (Humboldt-Universität Berlin)
Abstract: Electricity load forecasts are in various ways valuable for the operation of utilities. However, for a sustainable risk management of utility operators not only a forecast of expected demand, but also knowledge about the uncertainty and dispersion of future load plays an important role. The aim of our research is to model and forecast generalized quantiles of electricity demand, which, in contrast to forecasts of the conditional mean, yield a whole picture of the distribution of electricity demand. We apply methods from functional data analysis to model dynamics of daily generalized quantile curves. Taking temporal dependence between curves into account allows us to conduct short term forecasts at an intraday resolution using multivariate time-series techniques.