基于属性值贡献度的K最近邻分类算法  

KNN classification algorithm based on attributes value contribution

作　　者：许燕青[1] 

机构地区：[1]闽南理工学院实践教学中心,福建石狮362700

出　　处：《宁德师范学院学报（自然科学版）》2017年第2期131-133,共3页Journal of Ningde Normal University(Natural Science)

摘　　要：提出一种改进的K最近邻(KNN)分类算法,首先对样本按类别进行聚类,计算未知样本与近邻样本间的平均相似度,接着考虑不同类别中近邻样本点的个数,最后得出未知样本与近邻样本间的各个属性值贡献度.实验结果表明,改进后的基于属性值贡献度的KNN分类算法提高了分类准确率.In this paper, an improved KNN classification algorithm is proposed. First, the samples are clustered by category, and the average similarity between unknown samples and neighboring samples is calculated. Then, the number of neighboring samples in different categories is considered. Finally, unknown samples and the contribution value of each attribute value between neighboring samples are worked out. The experimental results show that the improved KNN classification algorithm based on the contribution of attribute value improves the classification accuracy.

关 键 词：KNN算法 分类 属性值 贡献度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]