Forecast Master Program case studies: Final report
- ed.
This report presents a number of case studies using the computer software package FORECAST MASTER (FM). The series studied and forecast are, aggregate monthly California Electricity Sales, system energy demand data from Ontario Hydro, peak demand data for the residential and commercial customers of Georgia Power Company, Massachusetts Electric commercial sales, Narragansett Electric commercial sales, average and peak demand using Georgia Power Company data. A variety of methods have been studied by each of the contributing authors; trend line fitting, exponential smoothing, Box-Jenkins univariate forecasting, vector autoregression, state space modeling, dynamic econometric models including time-varying parameters and general order serial correlation corrections. Thus both the data sets and the modeling/forecasting methodologies are varied. A number of conclusions emerge from these case studies: FM provides a powerful set of tools to aid a utility forecaster, a great deal of caution should be exercised in pre-processing the data; it can have unintended side effects, diagnostic tests are very useful in econometric models, the Akaike Information Criterion is a useful measure for selecting the best state space model, and state space and econometric approaches both need equal amounts of care in model analysis and presentation.
- Research Organization:
- Quantitative Economic Research, Inc., San Diego, CA (USA); Electric Power Research Inst., Palo Alto, CA (USA)
- OSTI ID:
- 6656412
- Report Number(s):
- EPRI-EM-5114; ON: TI87920354
- Country of Publication:
- United States
- Language:
- English
Similar Records
Forecasted electric power demands for the Delmarva Power and Light Company. Volume 1 and Volume 2. Documentation manual
Incorporating DSM impacts in econometric models: Results from a recent model enhancement study by the New York State Electric & Gas Corporation