基于决策级融合的成绩预测方法研究  

Research on performance prediction method based on decision level fusion

作　　者：李劲松 王娜 姚明海[1] 刘鸿雁 张野[1] LI Jin-song;WANG Na;YAO Ming-hai;LIU Hong-yan;ZHANG Ye(School of Information Science and Technology,Bohai University,Jinzhou 121013,Liaoning Province,China;School of Education Science,Bohai University,Jinzhou 121013,Liaoning Province,China)

机构地区：[1]渤海大学信息科学与技术学院,辽宁锦州121013 [2]渤海大学教育科学学院,辽宁锦州121013

出　　处：《信息技术》2024年第9期78-83,共6页Information Technology

基　　金：辽宁省社会科学规划基金项目(L22BTJ002);渤海大学博士启动项目(0519bs016);渤海大学校级教学改革研究项目(BHUXJGZ2022150);渤海大学校级科研项目(0524xn03804)。

摘　　要：针对当前基于单一方法构建的预测模型普遍存在预测准确性较低、泛化能力较弱的问题,提出基于决策级融合的成绩预测方法。首先,分别构建基于高斯过程回归和偏最小二乘的成绩预测模型;然后,根据两个模型的预测结果调整各自权重;最后,将两个模型的决策加权融合得到最终的预测结果。为了验证提出方法的有效性和稳定性,在某校化学、汉语言文学等七个专业真实数据上进行了大量的随机实验,并同主流预测方法进行对比。实验结果表明,提出方法具有更高的预测性和稳定性,可以为师生改进教学方式提供更为可信的决策支撑。To solve the problems of low prediction accuracy and weak generalization ability of prediction model based on single method,a performance prediction method based on decision level fusion is proposed.Firstly,performance prediction models based on Gaussian process regression and partial least squares are constructed,respectively.Then,the weights of the two models are adjusted according to the prediction results.Finally,the final prediction result is obtained by combining the decision weights of the two models.In order to verify the effectiveness and stability of the proposed method,a large number of random experiments are carried out on the real data of seven majors such as Chemistry and Chinese Language and Literature in a university,the reaults which are compared with the mainstream prediction methods.The experiment results show that the proposed method has higher prediction performance and stability,and can provide more credible decision support for teachers and students to improve teaching and learning methods.

关 键 词：教育数据挖掘 特征选择 学位预测 偏最小二乘回归 高斯过程回归 

分 类 号：TP301[自动化与计算机技术—计算机系统结构] G420[自动化与计算机技术—计算机科学与技术]