改进的BP神经网络在分数线预测中的应用

摘要

In order to improve the accuracy of pass mark predication,the paper presents an improved back-propagation neural network forecasting model which is consistent with the variation of pass marks on the college entrance examinations.An analysis is made of the main factors influencing pass marks on college entrance examinations,and the principal component analysis is used to reduce the dimension of influencing factors and normalize the data.And then an adaptive neural network is established to calculate the optimal weights by using the back-propagation algorithm.The best activation function and node number selection are given.Finally the improved model is applied to predict the pass marks on the college entrance examination for Xi’an Technological University.Results show that the neural network model can effectively improves the accuracy of pass mark prediction,providing a reference for candidates in the application process.%为了提高当前分数线预测模型的预测精度,提出一种符合分数线变化规律的反向传播神经网络分数线预测模型,并对其进行改进.文中分析了高考分数线的主要影响因素,运用主成分分析法对影响因素降维并归一化数据,建立分数线神经网络并改进其节点难以自适应的问题,得出最佳激活函数及节点个数选择,利用反向传播算法作为最终的学习算法计算出网络的最优权值.将建立的模型运用到西安工业大学高考录取分数线的预测上.结果表明,该分数线神经网络模型有效地提高了分数线预测精度,为高考分数线的预测以及学生志愿的填报提供了参考依据.