基于AHP和CRITIC综合赋权地K-means算法①.PDF

计 算 机 系 统 应 用 2016 年 第 25 卷 第 7 期 ① 基于AHP 和CRITIC 综合赋权的K-means 算法 丁晓琴, 张德生 (西安理工大学 理学院, 西安 710054) 摘 要: 传统的K-means 算法认为被分析样本的各个属性在聚类中作用是相同, 针对这种不足, 提出一种基于AHP 和 CRITIC 综合赋权的 K-means 聚类算法. 首先利用 CV-K-means 方法计算每个属性的权重, 从而两两进行比较得 到判断矩阵. 然后, 根据层次分析法(AHP)确定各个属性的主观权重, 再利用CRITIC 方法确定各个属性的客观权重. 采用差异系数法确定组合系数, 实验证明该算法的聚类精确度高于传统 K-means 算法. 关键词: K-means; 属性权重; AHP; CRITIC K-Means Algorithm Based on Synthetic Weighting of AHP and CRITIC DING Xiao-Qin, ZHANG De-Sheng (College of Science, Xi’an University of Technology, Xi’an 710054, China) Abstract: The traditional K-means algorithm is regarded that the attributes of swatches have the same effect on the clustering analysis. Based on AHP and CRITIC, comprehensive weighting of K-means clustering algorithm is proposed to solve the problem in this paper. First, each of attribute weight is calculated by CV-K-means method, thus judgment matrix is determined by comparing the two.Then, according to the analytic hierarchy process subjective weights of attributes is determined. And using the CRITIC method the objective weight of each attribute is determined, difference coefficient method is used to determine coefficient of combination. The experimental results show that the algorithm accuracy is higher than the traditional K-means algorithm. Key words: K-means; weight; analytic hierarchy process; criteria importance through intercriteria correlation 聚类(Clustering)是数据挖掘的一种重要技术, 是 倪少凯(2002)[1]为了更好融合各种赋权法的优点, 采用 分析数据并从中发现有用信息的一种有效的手段. 聚 不同类的赋权方法进行组合. 上官延华等人(2010)[2]将 类属于一种无监督的学习方法. K-means 算法是较为 客观权法中的熵值法和均方差法相结合, 得出综合赋 常用的聚类方法之一, K-means 算法将欧氏距离作为 权法K-means 算法