A Clustering Analysis Method With High Reliability Based on Wilcoxon-Mann-Whitney Testing

As a core step in clustering analysis, distance measurement results can influence accuracy. Existing methods are mostly based on cluster feature information. However, these features may be insufficient and result losing data information for clusters containing number of objects. To improve accuracy, we make full use the distribution characteristics objects clusters, i.e., descriptive statistics Wilcoxon-Mann-Whitney rank sum test nonparametric to measure distances during clustering. Furthermore, propose two-stage algorithm analysis performance. In terms avoiding preliminarily assuming with proposed method, discover arbitrary shapes Experiments multiple datasets compared other algorithms illustrate accuracy efficiency algorithm.