Gold Price Forecasting Based on Projection Pursuit and Neural Network

Abstract

Gold price has significant nonlinear, time-varying, many influence factors and more difficult to determine. In order to improve the precision of forecast gold price, this paper proposes a gold price combination forecasting model based on pursuit algorithm and neural network. Firstly, projection pursuit algorithm is used to choose the influence factors, and then the selected impact factors are used as BP neural network input variables to learn and establish gold price forecast model, finally forecast performance is tested by simulation experiments. The test results show that combined model can well depict the gold price trend, simplify network structure, accelerate the network learning speed, and improve the prediction accuracy of gold price. It provides a new forecast method for gold price.