摘要
近来,关于全面推进家庭经济困难学生认定工作精准资助为高校学生资助体系构建提供了一个崭新视角。如何更好的利用学生信息完成家庭经济困难精准认定工作是文章重点工作。文章以机器学习为基础,将学生信息库进行清洗,利用基于差分进化的特征选择为数据进行预处理,去除冗余特征,降低数据维度,以2个标准数据集与1个采集数据集对特征选择结果在2个分类器上进行有效性验证。以近2000名学生的信息为数据样本,通过K近邻分类预测算法预测学生家庭经济困难程度,验证了算法的可行性以及准确性。为大数据在高校教育中的应用提供了新的模式和方法。
Recently,the comprehensive promotion of family financial difficulties students to identify the work of precision funding for colleges and universities has provided a new perspective for the construction of student funding system.How to make better use of student information to complete the accurate identification of family financial difficulties is the key work of this paper.Based on machine learning,the student information base is cleaned,and the feature selection based on differential evolution is used to preprocess the data to remove redundant features and reduce the data dimension.The validity of feature selection results on two classifiers is verified by two standard data sets and one acquisition data set.Based on the information of nearly 2000 students as data samples,the K nearest neighbor classification and prediction algorithm is used to predict the economic difficulties of students'families,and the feasibility and accuracy of the algorithm are verified.It provides a new model and method for the application of big data in college education.
出处
《高教学刊》
2021年第3期76-79,83,共5页
Journal of Higher Education
基金
黑龙江省2017年高等教育教学改革研究项目“内涵式STAR研究生复试体系研究与实践”(编号:SJGY20170512)。
关键词
精准资助
大数据
差分进化
特征选择
K近邻预测
accurate funding
big data
differential evolution
feature selection
K neighbor prediction
