Bytkin S., Ivanyutin I., Kolomiiets V., Mokiy A. Steel price forecasting using the combination of methods of correlation and regression analysis and exponential smoothing of time series among STATISTICA 8.

Abstract

It is executed the analysis of dependence of cost of thin-sheet rolled metal from the value of raw materials with the use of applied software of «Statistica-8.0». The worked out method can be offered for prognostication of cost of rolled metal in the world market with probability more than 89 %. Timely information on trends in the markets plays an important role in maintaining the competitiveness of the company and as a result, their economic security. Proposed a multiple regression model, depending on metal prices of raw materials used in its manufacture. The model developed with sufficient accuracy reflects the price situation on the market. However, in practice there are discrepancies between the calculated and published forecast data. Using a combination of methods of regression analysis and exponential smoothing time series allows you to get close to the actual value of metal prices. This confirms the ability to use both computational and logical-heuristic methods of forecasting. In the case that the numerical values of metal prices, significantly reduces the risk of product delivery at economically unreasonable prices