摘要
目前关于股票聚类方法的研究比较多,但针对聚类结果优劣的评价尚无统一标准。针对股票财务聚类要求准确性和面向投资者的实用性等特点,将Ward聚类法的核心思想和信息熵理论相结合,基于偏差和损失最小与信息量损失最小两个角度,提出评价股票聚类结果优劣的Ward权熵指标,以验证在聚合聚类的条件下该指标关于聚类数K单调不降。
There are many ways of stock clustering while few methods about the evaluation of stock clusters .According to the specificity of stocks clustering , an index called Ward‐weighted entropy is introduced to evaluate the validity of clustering .This index bases on both the minimum of deviation loss and the amount of information loss .It can be proved that for agglomerative clustering ,the Ward‐weighted entropy index is non‐decreasing of the clustering number .Also the empirical analysis is introduced to show the features of this index ,and applying the index ,different clusters can be evaluated .
出处
《统计与信息论坛》
CSSCI
北大核心
2015年第1期29-34,共6页
Journal of Statistics and Information
关键词
股票聚类评价
Ward聚类法
信息熵
Ward权熵指标
stock clustering evaluation
Ward'
method
information entropy
Ward-weighted entropy indicator