A Probabilistic Forecasting Methodology Applied to Electric Energy Consumption
Probabilistic forecasting is presented as a technique required when the period to be described is in the far future and the relationships are complex. Some inputs to the forecasting model may be precise and objective, but these must be suitably related to others that are based on subjective estimates, educated guesses at best. The example given is the probabilistic forecast of U.S. demand for electricity in the year 2000. The results suggest a consumption between 3500 and 4500 billion kWh, only about half the consumption predicted by extrapolating the compound growth trends of recent history. With the probabilistic — as opposed to the deterministic — forecast, the relative risks of alternative plans can be quantified and can thus be converted into decision criteria for planning and for policy comparisons.