K-means算法在玉米良种选育中的应用  

K-means algorithm application in maize seed breeding

作　　者：杨娜[1] 邱建林[2] 潘阳[1] 卞彩峰 陆鹏程[1] 

机构地区：[1]南通大学电子信息学院,江苏南通226019 [2]南通大学计算机科学与技术学院,江苏南通226019

出　　处：《计算机工程与设计》2014年第8期2864-2871,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(NSF61272424);江苏省自然科学基金项目(BK2010277);南通市科技计划基金项目(K2010002;AL2007033)

摘　　要：为降低k值选取对聚类效果的影响,用改进的聚类有效性函数来选取k值。综合考虑了类内相似性、类间差异性,通过UCI数据库中的几个数据集和k-means算法来验证其正确性,并应用于玉米良种选育。为确保得到的玉米良种集中不混有劣种,对聚类得到的玉米良种集进行主成分分析和熵值法的综合评价。实验结果表明,采用该函数在玉米良种集中得到了优良的玉米种子,验证了其正确性。To reduce the effect of the k value selection on clustering results, the improved clustering validity function was used to select the k value. The similarity within the class and differences between classes were taken into consideration. The correctness of the function was verified by several datasets of the UCI database and the k-means algorithm. Finally the function was applied in the maize seed breeding. To ensure the fine corn set without mixing with inferior seeds, the comprehensive evaluation by the principal component analysis and the entropy value method was made. Experiments prove the validity of the improved clustering validity function, and good corn seeds were gotten in maize seed breeding.

关 键 词：K-MEANS算法 主成分分析 熵值法 聚类有效性函数 玉米良种选育 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]