نوع مقاله : پژوهشی
عنوان مقاله English
نویسندگان English
Today's world's dependence on technology increases human need for products produced from iron ore, and predictions indicate that steel demand will increase by 60 percent by 2035 (Mohammadi, Soltani Mohammadi, and Bakhshandeh Omnieh, 2013). For this reason, forecasting the price of metals, including iron ore, using quantitative and qualitative methods such as technical and economic market studies has not been very consistent with reality. One of the common methods of examining prices is the time series method. In this study, iron ore prices were predicted by modeling and using time series analysis with the help of a dynamic neural network. Next, the price of iron ore was estimated by using an artificial neural network method and considering the monthly price of iron ore and the factors affecting its fluctuations, and then the results obtained were evaluated in terms of predictability. The optimal neural network model with 3 layers and 10 neurons has estimated the price of iron ore with very good accuracy. In this model, the training error is about 1.7%, and for validation it is 2.3%, and the test error is 1.5%. Also, the regression and correlation values of the data at a confidence level of 95% and a high correlation value of R2=0.98 indicate a good model with good accuracy.
کلیدواژهها English
Copyright ©, Pourya Farajian, Nima Farajian
License
This article is released under the Creative Commons Attribution (CC BY 4.0) license. Anyone is free to copy, share, translate, and adapt this article for any purpose, whether commercial or non-commercial, as long as proper citation is given to the authors and original publication.
