基于改进的BP网络对CET-4累计通过率时间序列的预测

时间序列并非总呈自相关性.与4]不同,对本文中的CET-4累计通过率时间序列建立线性及二阶自回归模型,仿真计算失效.改进的BP网络,适用所有的一维时间序列.本文采用二步预测法,与其它采用BP网络对时间序列预测不同的是,本文不仅预测下一年的时间序列值,还将整个预测模型仿真出来,画出三维图形,从而为教务政策的制订提供直观易看、合理、客观的依据.

The Application of Improved BP Neural Network in the Prediction of the Accumulative Total Passing Rates of CET-4 Time Series

Time series are not always autocorrelation.The CET-4 time series accumulative total passing rates of this paper is distinct from .The autoregressive model and the quadric polynomial model are refused by simulated computing.The improved BP neural network model can apply in all one-dimension time series.The two steps method of prediction is employed.Unlike other paper,this paper not only gives the prediction time series value of the next year,but also gives the whole predict model,and gives the 3-dimensional figure.It provided a visual and reasonable reference for the policies' setting down of the management of the educational administration.