基于DNN的大学生学业成绩预测方法研究——以北京市某高校电子信息类专业为例

Research on Prediction Method of College Students Academic Achievement Based on DNN——Take the Electronic Information Major of a University in Beijing as an Example

本文以北京市某高校电子信息类专业2011—2015级学生的8万多条成绩数据为基础,针对学生所学课程相关性和成绩预测展开研究。首先利用关联规则和决策树组合算法,对课程之间的关联性进行深度挖掘,扩宽分析预测结果的覆盖面。在此基础上,构建基于深度神经网络(Deep Neural Network,DNN)模型的学生成绩预测方法,以学生低年级(大一大二)的已修课成绩来对高年级(大三大四)的未修课成绩进行预测,实现成绩预警功能,同时将预测结果应用于教育教学中,有利于推动课程的优化设置,进而实现提高育人质量的目标。

Based on more than 80,000 pieces of grading data of the students majoring in Electronic Engineering in one University in Beijing,this paper focuses on the research of courses relationship mining and courses grades prediction.Firstly,the combination algorithm of association rule and decision tree has been proposed,which is used to mine the course inner relationship and widen the coverage of courses grades prediction.Then the Deep Neural Network(DNN)has been implemented to leaning the mapping between lower-division courses(courses of freshman and sophomore students)and higher-division courses(courses of junior and senior students).This DNN grades prediction model can be used for predicting the higher division courses grades when input the lower division courses grades,which can also be used for students’academic performance alerts.The models of this paper can be used to promote the optimization of the curriculum sets,thereby achieving the goal of improving the quality of higher education.