It has been shown that the method has more accurate forecasts,reinforcement learning properties and mapping capabilities by comparing with LPM,Fisher model and Logistic model.

• 通过与多元线性回归模型、Fisher模型和Logistic回归模型的预测结果对比表明,该方法具有预测精度高,学习与泛化能力强,适应性广的优点。