A family of techniques each of which aims to estimate some aspect of the future.
Depending on the subject under review, the future may be regarded as either non-controllable or controllable, e.g. the weather is non-controllable, personal spending on luxuries is controllable, future sales are partially controllable. The two main categories lead to two branches of forecasting, exploratory and normative, each with their own techniques. Taking the latter first, if the future is controllable, forecasting becomes a matter of specifying the desired or intended future and becomes in practice, inseparable from planning (q.v.). Where the future is non-controllable you use methods such as extrapolating past trends, or finding indicators that are related to a variable in which you are interested but with a lead or lag (unemployment and the level of orders for machine tools for example). Commercially, for sales forecasting and other purposes the main techniques used are analysis of time series (q.v.), regression analysis, exponential smoothing (q.v.), and the use of economic indicators. But there are several hundred detailed techniques available including much more unusual examples such as the Delphi technique, scenario writing and contextual mapping. The great majority of techniques are described in detail by Jantsch.
Examples of practical applications are given under the detailed techniques referred to above. Hermann Khan has used sophisticated techniques in great depth in his American think-tanks to arrive at forecasts for life beyond the year 2000. The NASA Apollo project is the classic example of maximum scale, maximum complexity normative forecasting.