我国大学生创业意愿识别模型比较研究  

A Comparative Study on the Recognition Models of Chinese College Students'Entrepreneurial Intention

作　　者：石峰[1] 胡燕 戴冬阳 SHI Feng;HU Yan;DAI Dong-yang(School of Management,Hunan Institute of Engineering,Xiangtan Hunan 411104;School of Law,Central South University,Changsha Hunan 410012;Department of Defense Economics,Army Logistics University of PLA,Chongqing 401311)

机构地区：[1]湖南工程学院管理学院,湖南湘潭411104 [2]中南大学法学院,湖南长沙410012 [3]中国人民解放军陆军勤务学院国防经济系,重庆401331

出　　处：《牡丹江大学学报》2021年第3期88-94,共7页Journal of Mudanjiang University

基　　金：湖南省教育科学规划项目“乡村振兴战略背景下大学生返乡创业意愿及影响因素研究”(XJK18BGD051)。

摘　　要：运用2015年中国综合社会调查(CGSS)数据,建立包括个人认知能力、个人特征、创业环境、社会信任和资源禀赋等五个维度的大学生创业意愿识别指标体系,构建逻辑回归、支持向量机、决策树和K最近邻算法的大学生创业意愿识别模型。四种算法模型的结果比较表明:基于10次重复试验的平均准确率排序依次为K最近邻﹥有序多分类逻辑回归﹥决策树的rpart﹥支持向量机。从Kappa系数平均值看,也得到与平均准确率一致的结论,即K最近邻模型的分类效果最好,支持向量机模型的分类效果最差。Using Chinese General Social Survey data of 2015,this paper establishes a five-dimensional index system for college students'entrepreneurial intention recognition including personal cognitive ability,personal characteristics,entrepreneurial environment,social trust and resource endowment,and construct logistic regression,support vector machines,Decision-making tree and K-nearest neighbor algorithm for college students'entrepreneurial intention recognition model.The comparison of the results of the four algorithm models shows that the ranking based on the average accuracy of 10 repeated trials is K nearest neighbor,ordered multiclass logistic regression,rpart of decision tree,and support vector machine.From the average value of Kappa coefficient,the conclusion that is consistent with the average accuracy is also obtained,that is,the classification result of the K nearest neighbor model is the best,and the classification result of the support vector machine model is the worst.

关 键 词：创业意愿 逻辑回归 支持向量机 决策树 K最近邻 

分 类 号：G640[文化科学—高等教育学]