基于偏最小二乘法的学生成绩预测研究  

Research on student achievement prediction based on partial least square method

作　　者：高艺璐 刘鸿雁 李劲松 潘子尧 GAO Yi-lu;LIU Hong-yan;LI Jin-song;PAN Zi-yao(College of Educational Science,Bohai University,Jinzhou 121013,Liaoning Province,China;College of Information Science and Technology,Bohai University,Jinzhou 121013,Liaoning Province,China;College of Mathematical Sciences,Bohai University,Jinzhou 121013,Liaoning Province,China)

机构地区：[1]渤海大学教育科学学院,辽宁锦州121013 [2]渤海大学信息科学与技术学院,辽宁锦州121013 [3]渤海大学数学科学学院,辽宁锦州121013

出　　处：《信息技术》2024年第6期47-53,共7页Information Technology

基　　金：辽宁省教育科学研究项目(WJ2020004,LJ2020003);辽宁省教学改革项目(10167-636,10167-649,10167-800);研究生教育教学改革项目(YJG20210002,YJG20210005);渤海大学本科教学改革研究项目(BHUXJGA2022030)。

摘　　要：成绩预测已经成为提高教学质量的重要手段之一。目前多数成绩预测方法都是单一的考虑课程成绩,而忽略了各门课程之间的相关性。考虑到课程之间的相互支撑关系,提出了基于偏最小二乘法的成绩预测方法,在传统成绩预测方法的基础上进一步充分考虑课程之间的相关性。为验证算法性能,以某校信息与计算科学专业的学生成绩为研究对象对算法进行了验证。实验结果表明,基于偏最小二乘法的成绩预测方法具有更低的预测误差,并且因其受训练样本数量影响较小,更易于泛化推广。Achievement prediction has become one of the important means to improve teaching quality.At present,most of the prediction methods only consider the course results and ignore the correlation between courses.Considering the mutual support between courses,a prediction method based on partial least square method is proposed,which further considers the correlation between courses on the basis of traditional prediction method.In order to verify the performance of the algorithm,the results of students majoring in information and computing science in a university are taken as the research object to verify the algorithm.The experiment results show that the performance prediction method based on partial least square method has lower prediction error,and it is easier to generalize because it is less affected by the number of training samples.

关 键 词：偏最小二乘法 成绩预测 教育数据挖掘 高等教育 显著性检验 

分 类 号：TP3-05[自动化与计算机技术—计算机科学与技术] G420[文化科学—课程与教学论]