Abstract:

Forecasting sales volume is an important element of a company’s operations. When forecast that are used for making further decisions turn out to be incorrect, a company incurs additional costs. Costs resulting from forecasting error may be caused by either over- or underestimation. In each of these cases, the costs may be different. Managers’ knowledge and intuition along with observations of the past allow for estimating the cost function, whose independent variable is the size of the error. This article proposes the forecast to be taken as the value from the given range which minimizes the costs of an incorrect decision. This approach is in accordance with the forecasting according to minimum loss rule. For solving this problem, Savage’s minimax regret criterium was used, generalizing it for continuous variables. The error cost was described by the use of an asymmetrical loss function, known from confidence theory.