特征融合和集成学习在大学生助学金预测中的应用  

Ensemble learning and feature integration in prediction for college students grant

作　　者：孙瑜[1,2] 李占利 李学文[2] SUN Yu;LI Zhan-li;LI Xue-wen(College of Computer Science and Engineering,Xi’an University of Science and Technology,Xi’an 710054,China;College of Safety Science and Engineering,Xi’an University of Science and Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学计算机科学与技术学院,陕西西安710054 [2]西安科技大学安全科学与工程学院,陕西西安710054

出　　处：《西安科技大学学报》2020年第4期744-750,共7页Journal of Xi’an University of Science and Technology

基　　金：陕西省联合基金项目(2019JLZ-08)。

摘　　要：随着数字化校园的推行,利用大数据和机器学习算法识别真正困难的大学生,实现助学金公平发放,可为高校资助问题提供技术支持和辅助决策。首先,提出了一种利用多特征融合和集成学习的助学金预测方法,对消费、成绩、出入宿舍、图书借阅等大学生日常行为数据进行预处理和特征提取,分析特征的重要性,并进行特征融合,构造了一个21维特征向量。然后,利用集成学习方法对梯度提升决策树,随机森林,AdaBoost,SVM等分类器进行集成,采用过采样和交叉验证的方法,利用不同组合策略对大学生助学金进行分类预测。通过对10885位大学生日常行为数据进行实验,结果表明,在3种性能指标(F1、召回率、精确度)上进行测试,平均精确度达到0.9545,为大学生助学金发放提供了一种辅助决策手段。With the implementation of digital campuses,big data and machine learning algorithms are used to exactly identify college students with financial difficulties,to realize the fair distribution of grant,it has provided new technical support and decision-making for university funding problems.First,a multi-classification method was proposed to predict the level of grant for each college student based on feature integration and ensemble learning.It extracted from expense,scores,in/out dormitory,book loan conditions of students’daily behaviors data and constructed a 21 dimensional feature.The ensemble learning method was used integrate gradient boosting decision tree,random forest,AdaBoost and support vector machine classifiers for college grant classification.The proposed method was evaluated with 10885 students set and experimental results show that the proposed method has a high average accuracy of 0.9545 and can be used as an effective means of assisting decision-making for college students grant.

关 键 词：集成学习 多类别分类 梯度提升决策树 大学生助学金 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]