RBF NEURAL NETWORK-BASED STOCKS MARKET MODELING AND FORECASTING
Zheng Pi · Journal of Tianjin University Science and Technology · 2000
An RBF neural network based method for stocks market modeling and forecasting is presented,and a hierarchical genetic algorithm is proposed to train network parameters such as RBF centers,widths,connection weights and the configuration.As a result,stocks marcket models to forecast the stocks marcket price and index are developed with RBF neural networks with well trained parameters and configuration,based on the real world data available from operation of stocks marckets.By forecasting the Shanghai stock market price index and the stock price of Yili it has shown that this method has reinforced learning properties and mapping capabilities.It is useful for modeling and forecasting of uncertain nonlinear systems.