论文部分内容阅读
针对股指预测的不确定性,提出一种以SARIMA模型的AR项和预测值作为RBF神经网络的输入变量,残差作为RBF的输出变量,建立SARIMA模型的新方法。实验研究结果表明这种股指预测方法可以得到较高的预测精度。
Aiming at the uncertainty of stock index forecast, this paper proposes a new method to establish SARIMA model by using the AR term and forecast value of SARIMA model as the input variables and RBF neural network output variables as RBF output variables. Experimental results show that this stock index prediction method can get a higher prediction accuracy.