| Summary: | Dissertation submitted for Masters of Science degree in Mathematical Statistics
in the
Faculty of Science,
School of Statistics and Actuarial Science,
University of the Witwatersrand
Johannesburg
May 2016 === Electricity demand in South Africa exhibit some randomness and has some important
implications on scheduling of generating capacity and maintenance plans. This work
focuses on the development of a short term probabilistic forecasting model for the 19:00
hours daily demand. The model incorporates deterministic influences such as; temperature
effects, maximum electricity demand, dummy variables which include the holiday
effects, weekly and monthly seasonal effects. A benchmark model is developed and an
out-of-sample comparison between the two models is undertaken. The study further assesses
the residual demand analysis for risk uncertainty. This information is important
to system operators and utility companies to determine the number of critical peak days
as well as scheduling load flow analysis and dispatching of electricity in South Africa.
Keywords: Semi-parametric additive model, generalized Pareto distribution, extreme
value mixture modelling, non stationary time series, electricity demand
|
|---|
