Yearly cluster analysis of Dow Jones 30 from 2006–2009 based on extreme point distance
Ling Xiang, Yun-Feng Chang, Yuanchao Liu, DaoChao Tang · 2012
In this paper, we propose to do emerging clustering analysis based on extreme point distance. In order to demonstrate the validity and feasibility of this method, analysis of the feature of Dow Jones 30 from 2006-2009 is done. The clustering results show that our clustering process could cluster stocks with similar evolution trend into the same cluster while dissimilar ones into different cluster, which confirm that the definition of extreme point distance is a good measure for stock distance. And the differences among the yearly clustering results show that these 30 stocks are independent from each other, which makes their focus change from year to year according to the evolution of market.