全局中心聚类算法在课程序化中的应用  

Application of Global Center Clustering Algorithm in Course Program

作　　者：段桂芹[1] 刘松[2] 邹臣嵩[3] DUAN Guiqin;LIU Song;ZOU Chensong(Department of Computer Science,Guangdong Songshan Polytechnic College,Shaoguan 512126;Department of Mechanical Engineering,Guangdong Songshan Polytechnic College,Shaoguan 512126;Department of Electrical Engineering,Guangdong Songshan Polytechnic College,Shaoguan 512126)

机构地区：[1]广东松山职业技术学院计算机系,韶关512126 [2]广东松山职业技术学院机械工程系,韶关512126 [3]广东松山职业技术学院电气工程系,韶关512126

出　　处：《计算机与数字工程》2020年第3期528-533,共6页Computer & Digital Engineering

基　　金：广东省科技厅科技发展专项资金(编号:2017A070712006);广东高校省级重大科研项目(编号:2017GkQNCX033);2015年度广东省高等职业教育专业教学标准研制项目(编号:BZ201511);韶关市科技计划项目(编号:2017CX/K055);广东松山职业技术学院重点科技项目(编号:2018KJZD001)资助。

摘　　要：针对K-means在聚类过程中存在的随机性强、准确率不稳定等问题,提出了一种改进聚类算法,首先选取k个首尾相连且距离乘积最大的数据对象作为初始聚类中心,在簇中心迭代过程中,选取簇内距离和最小的样本作为簇中心,再将其他样本划分至相应簇中,反复迭代,直至收敛。在UCI数据集上的仿真实验结果表明:新算法与K-means算法和其他两种改进算法相比,不仅能够降低运算耗时,在准确率、Jaccard系数、F值等多项聚类指标上也有较大的提升,在实际应用中,使用新算法对现代学徒制的职业能力进行了聚类分析,解决了课程间的序化问题。In order to solve the problems of strong randomness and unstable accuracy in K-means clustering process,an improved clustering algorithm is proposed. Firstly,k data objects connected to each other and with the largest distance product are selected as the initial clustering centers. In the process of updating the cluster center,the data object with the smallest sum of the sample distance in the cluster is selected as the cluster center,and the other data objects are divided into the corresponding clusters according to the minimum distance,and iterated repeatedly until convergence. Through the simulation experiment in UCI data set,the clustering performance of K-means algorithm and other improved clustering algorithms are compared. The results show that the new algorithm can not only reduce the computation time of the algorithm. There is also a great improvement in many clustering indexes such as accuracy rate and Jaccard coefficient F value. In practical application,the new algorithm is used to analyze the related data of the course system in order to solve the problem of ordering between courses.

关 键 词：全局中心 簇内距离和 序列聚类 课程体系 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]