Modeling electricity loads in California: ARMA models with hyperbolic noise

Authors:
J. Nowicka-Zagrajek;R. Weron
Affiliations:
Institute of Mathematics, Wroclaw University of Technology, 50-370 Wroclaw, Poland;Hugo Steinhaus Center for Stochastic Methods, Wroclaw University of Technology, 50-370 Wroclaw, Poland
Venue:
Signal Processing - Signal processing with heavy-tailed models
Year:
2002

Citing 4
Cited 1

Time series: theory and methods

Time series: theory and methods
The Non-Gaussian nature of bibliometric and scientometric distributions: a new approach to interpretation

Journal of the American Society for Information Science and Technology
Linear and nonlinear time series forecasting with artificial neural networks

Linear and nonlinear time series forecasting with artificial neural networks
Stock returns and hyperbolic distributions

Mathematical and Computer Modelling: An International Journal

Forecasting electricity demand in Japan: A Bayesian spatial autoregressive ARMA approach

Computational Statistics & Data Analysis

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we address the issue of modeling and forecasting electricity loads. We apply a two-step procedure to a series of system-wide loads from the California power market. First, we remove the weekly and annual seasonalities. Then, after analyzing properties of the deseasonalized data we fit an autoregressive moving average model. The obtained residuals seem to be independent but with tails heavier than Gaussian. It turns out that the hyperbolic distribution provides an excellent fit. As a justification for our approach we supply out-of-sample forecasts. As it turns out, our method performs significantly better than the one used by the California System Operator.