摘要
为了提高大学生学业状态的预警准确性,在经典的SMV模型的基础上,调整核参数以及性能被惩罚的参数值,创建了改进的BAS-SVM的大学生学业状态动态预警模型。为准确合理地实现学业动态预警,选择考勤指数、学习指数、成绩指数、“宅”指数和家境等5个关键指标作为学业预警的主要指标,并将学业预警结果划分为好、中、差三个等级。将影响学业状态的5个关键BAS-SVM模型的输入变量为影响学业状态的5个关键KPI值,经过模型运算后输出预警判断结果值。与ELM、SVM和BPNN对比发现,BAS-SVM可以有效提高学业状态预警结果的准确率,为学业状态预警评估提供了新的方法和途径。
The performance of SVM model is influenced by penalty parameter and kernel parameter.In order to improve the early warning accuracy of college students'academic state,this paper proposes a dynamic early-warning model for college students'academic status based on BAS-SVM.In order to accurately and reasonably realize the dynamic early warning of school work,five key indicators,such as attendance index,study index,achievement index,dwelling index and family condition,are selected as the main indicators of the early warning of school work.The results of academic early warning are divided into good,medium and poor grades.Five key indicators affecting academic status are taken as the input of BAS-SVM,and the results of academic early-warning are taken as the output of BAS-SVM.Compared with ELM,SVM and BPNN,BAS-SVM can effectively improve the accuracy of early warning results of academic status,and provide a new method and approach for early warning assessment of academic status.
作者
卢毅
LU Yi(School of Economics and Management,Shanxi Xueqian Normal University,Xi’an 710061,China)
出处
《微型电脑应用》
2021年第1期111-114,共4页
Microcomputer Applications
关键词
支持向量机
学业状态预警
天牛须搜索算法
决策树
关联规则
support vector machine
early warning of academic status
Beetle antennae search algorithm
decision tree
association rules