CLUSTERING BASED ON BETWEENNESS CENTRALITY IN PERIOD: TRANSFORMATION OF CORRELATION COEFFICIENT VALUE INTO DISTANCE IN MATRIC SPACE

The main information of this research is the transformation of the correlation coefficient value for stock price into the distance. It is done to create a representation in metric space that can be used in cluster analysis on the correlation network, which is a dynamic network. The dynamic network is generated from the weighted edges in the form of distances in each period. Finding the cluster members of the network can be analyzed using a simple technique called a minimum spanning tree. The central cluster member is the vertex betweenness. Vertex betweenness represents banking companies with a high degree of proximity and correlation. It means that the banks that are members of the central cluster are banks with high investment value. Clustering based on betweenness centrality in the case study of stock price correlation becomes useful when transforming the value of the correlation coefficient to distance. The effort to build a network with the edge weight being the distance makes the minimum spanning tree a simple, valuable method for cluster analysis on bank stock prices. In particular, the benefit to investors, i.e., it can reveal which assets are closely correlated, indicating that they may respond to market events in a similar way and make decisions in stock purchases