【摘要】 由于学生英语写作成绩预测受诸多因素影响,具有高维、非线性特点,本文基于广义回归神经网络(GRNN)算法原理,构建了GRNN学生英语写作成绩预测模型,并与弹性BP算法改进的BP神经网络模型的预测结果进行对比分析。仿真结果表明:改进的BP神经网络模型的预测最大相对误差为3.23%,GRNN模型的预测最大相对误差仅为-0.72%,表明所建立的GRNN模型的预测精度高、泛化能力强、收敛速度快、调整参数少,验证了将GRNN应用于英语写作成绩预测方案的可行性。更多还原

【Abstract】 Prediction of student English writing scores is influenced by various factors. It has high dimensional and nonlinear features. A prediction model of students’ English writing scores was established in this paper. The model was based on the algorithm principle of generalized regression neural networks(GRNN). Its prediction result was analyzed and compared with that of a BP neural network model improved by resilient back-propagation. The simulation results indicate that: the largest relative error of prediction produced by improved BP neural network model is 3.23%, while the one produced by GRNN model is only-0.72%. This implies that the GRNN model has higher prediction accuracy, better generalization ability, faster convergence speed and less adjusting parameters. Thus the feasibility of applying GRNN to English writing score prediction is verified.更多还原