Stochastic time-series modelling for long-term load forecasting

Sometime series models suitable for forecasting are reviewed. Autoregreasive moving average-type (ARMA) time series models, in particular, are well suited for forecasting applications. The procedure of model development, consisting of the identification, estimation, and diagnostic checking stages, make these models convenient to apply to practice. The accuracy of load forecasting can-be improved by increasing the number of observations. Application of these models to Egyptian electricity consumption is presented, and different model forecasts are compared.