Electricity authorities in the UAE have not been successful in develop
ing reliable and accurate models of system peak load. In this study, w
e develop a time-series-based decision-support system that integrates
data management, model base management, simulation, graphic display, a
nd statistical analysis to provide near-optimal forecasting models. Th
e model base includes a variety of time-series techniques, such as exp
onential smoothing, Box-Jenkins (BJ), and dynamic regression. The syst
em produces short-term forecasts (one year ahead) by analyzing the beh
avior of monthly peak loads. The performance of the DSS is validated t
hrough a comparison with results suggested by econometricians. (C) 199
7 Elsevier Science Ltd.