Lin Yongxing; Si Yanru
Abstract
One of the most widely used metals in the world is the Iron. The world cost of iron ore is defineded by its supply and demand. Numerous variabes such as steel, scrap, oil, gold, interest rate, inflation rate, dollar value, and stock value affect the world price of iron ore. Therefore, for economic investment ...
Read More
One of the most widely used metals in the world is the Iron. The world cost of iron ore is defineded by its supply and demand. Numerous variabes such as steel, scrap, oil, gold, interest rate, inflation rate, dollar value, and stock value affect the world price of iron ore. Therefore, for economic investment of iron ore, it should be forecasted precisely by the scientists to give a direction to the decision makers to make a proper decision for the society. Due to the multiplicity of effective parameters and the complexity of the relationships between the iron ore variables, artificial intelligence is the best idea for forecasting. In this paper, we utilized a new optimized version of Convolutional Neural Network (CNN) to facilitate this task. To do so, a modified version of the Search and Rescue (MSAR) optimization algorithm has been designed and used for optimizing the CNN for improving its training efficiency in forecasting the iron ore price volatilities. The method is then validated based on ten different variables. Finally, a comparison of the results with various state of the art techniques was carried out to show the suggested method effectiveness. The results showed that the suggested technique has the fittest results in comparison to the other newest techniques.
