Rare Earth Price Fluctuation and Forecasting Methods under the COVID-19 Pandemic

Abstract

As an important strategic resource, rare earth price has its fluctuation rules under the COVID-19 Pandemic and its price prediction are of great significance to the increase of mineral benefits in the future. A BP neural network (ACO-BP) combination model based on Ant Colony Optimization was constructed to predict the price of rare earth products in terms of the factors that affecting the price of rare earth resources. Principal component analysis is used to eliminate redundant information among influencing factors, which can reduce the input data dimension of BP neural network and improve the prediction accuracy. Then the Ant Colony algorithm is used to find the optimal neural network threshold to optimize the convergence rate of the model and reduce the prediction error. Taking dysprosium oxide price as sample, monthly data from January 2010 to March 2018 are selected to construct a multi-factor ACO-BP combination model for prediction. The results show that the ACO-BP combined model is superior to the traditional BP neural network model in simulation ability, error level and convergence accuracy, and can predict the dysprosium oxide price more accurately. This method may benefits for Rare Earth industry after the COVID-19 epidemic.