In order to get the excellent accuracy for price forecast in the steel market, the adaptive Radial Basis Function (RBF) Neural Network (NN) model, Back Propagation (BP) NN model and Sliding Window (SW) model are utilized to forecast the price of the steel products in this paper. Eight steel products, which extracted from Shanghai Baoshan steel market of China at January, 2011 to December 2011, are selected to forecast the price about one week and compare the Mean Absolute Errors (MAE) by RBF model, BP model and ASW model respectively. One main parameter of each model's is changed step size by programs automatically. Experiments demonstrate that the ASW model is best model which can get lowest Mean Absolute Errors (MAE). Experiment results prove that the proposed ASW model is meaningful and useful to analyze and to research the price forecast in the steel products market.
展开▼
