Uwagi do metodyki prognoz frachtowych

Abstract

Some remarks on the methodology of freight forecasts are presented in the article. The problem of choise of the best method of freight forecasts is the subject of this study. Linear regression functions of nine and eight variables and some methods of analysis of time series of freight indexes are considered, and among them — a method of harmonic weights, exponential smoothing and adaptive forecasting. It is shown, that both the methods have to be used because, regression functions better explain the mechanism of freight rates and they give better forecasts of turning points. Adaptive forecasting, with parametres α=0,7; ß=0,38, give the best forecasts among the methods of trend analysis.