T-平方统计在类可聚性检验中的应用

Application of T-Square Statistic to Tests for Clustering Possibility

  • 摘要: 本文讨论T-平方统计在模式类可聚性检验中的应用问题.实践证明,在应用聚类算法之前确定模式集合是否存在聚类结构,可以帮助我们避免对聚类算法的不适当应用,从而有效化聚类结果.Monte-Carlo实验和对IRIS等数据的实际检测指出,T-平方统计结合半数框架应用,对良好聚类的模式集合有较强的检验功效.

     

    Abstract: This paper discusses the problem of application of T-square statistic to tests for clustering possibility.It is proven by practice that determining if there is some clustering structure in the data set before applying any clustering algorithm can help us to avoid improper application of the algorithm and validating the results.Monte Carlo experiment and actual tests for IRIS as well as another data set indicate that T-square statistic applied with the half number frame has high detection power for well clustered data set.

     

/

返回文章
返回