Method for determination of the predicted probabilities of events when making decisions

In this paper, we developed a method to estimate the probability of occurrence of the forecasted events. The method uses a priori data about the prediction of relevant events in the past periods and data of observed events at present. For both types of data there are generated two matrices, one of which describes the prediction error known from previous periods, and the other contains detailed estimates obtained on the basis of new information obtained at present. The product of these matrices forms a complete matrix of prediction errors, which characterizes the complete error of subject, or the expert when making forecasts. It is shown that the vector of probabilities of occurrence of the forecasted events is the eigenvector of the full matrix prediction error, which is responsible to the unit eigenvalue of this matrix. While in the Bayesian approach a posteriori probabilities of forecast events defined by their a priori probabilities, which, in principle, are not known a priori, in the developing method the predicted probabilities of events are unique determined from the received equations. Application of the method is considered on the example of prediction future demand for the new product, in order to make sciencebased decisions on its production.