红斑鳞状皮肤病的聚类分析

发布时间：2018-06-01 23:47

本文选题：红斑鳞状皮肤病 + k-均值算法　；参考：《济南大学学报(自然科学版)》2017年03期

【摘要】：针对红斑鳞状皮肤病鉴别诊断难题,提出利用聚类分析进行诊断;采用3种k-均值、2种k-中心点、最小生成树以及密度峰值点快速搜索聚类算法对该疾病数据进行分析,比较各算法对该疾病的聚类误差平方和、聚类结果 Rand指数、Jaccard系数、调整Rand指数以及聚类准确率;比较各算法对预处理的数据子集的聚类结果与文献中采用k-均值算法对未预处理的该数据子集的聚类结果。结果表明:邻域k-中心点算法对红斑鳞状皮肤病有很好的聚类效果,聚类准确率、聚类结果 Rand指数、Jaccard系数、调整Rand指数均优于对比算法,密度全局k-均值算法的聚类效果次之,全局k-均值算法取得最佳聚类误差平方和;k-均值算法对预处理数据子集的聚类准确率最高,邻域k-中心点与密度全局k-均值算法的聚类准确率相等;数据预处理可提高k-均值算法对该疾病的诊断准确率。
[Abstract]:Aiming at the difficult problem of differential diagnosis of erythema squamous dermatosis, a cluster analysis method is proposed, and three kinds of k-mean points, two kinds of k-center points, the minimum spanning tree and the fast searching clustering algorithm of density peak point are used to analyze the data of the disease. The clustering error square sum of each algorithm for the disease is compared, the Rand index and the Jaccard coefficient of the clustering result are compared, and the Rand index and the clustering accuracy are adjusted. The clustering results of the pre-processed data subsets are compared with those of the unpreprocessed data subsets by using the k-means algorithm in the literature. The results show that the neighborhood k- center algorithm has a good clustering effect on erythema squamous dermatosis, the clustering accuracy, the clustering result Rand index and the adjustment of Rand index are better than the contrast algorithm. The clustering effect of the density global kmean algorithm is the second, the global kmean algorithm obtains the best clustering error square sum, and the K-means algorithm has the highest clustering accuracy for the pre-processed data subset. The clustering accuracy of the neighborhood k- center algorithm is equal to that of the density global k- mean algorithm, and the data preprocessing can improve the diagnostic accuracy of the disease.
【作者单位】：陕西师范大学计算机科学学院;
【基金】：国家自然科学基金项目(61673251) 陕西省科技攻关项目(2013K12-03-24) 中央高校基本科研业务费专项资金项目(GK201503067)
【分类号】：R758.6;TP311.13

【相似文献】