Hybrid Method For Forecasting A Steel Mill's Sales

Abstract

This paper presents a proposal of long range steel mill's sales forecasting method. It can be also used in other sectors of steel industry, and (after minor modifications) in enterprises operating in a power industry, building materials industry, plastics industry etc. Heuristic methods are often used in long range forecasting. In the proposed method few parameters are predicted with fuzzy Delphi method. It results in the forecasts in form of fuzzy triangular numbers. Transformation of possibility distribution, generated by fuzzy number, into a probability distribution enables to integrate forecasts obtained from fuzzy Delphi method with those prepared with structural models and developmental trends models. A computer simulation is used to determine sales forecast's expected value and probability distribution. In case of structural models and developmental trends models a simulation with re-estimation has been used. The method presented in this paper is used to generate random numbers from a set determined by triangular fuzzy number.