Taxonomy Of The Stocks

This thesis employs the minimum spanning tree method to study the classification of the stocks. Firstly, the correlation coefficients for the stocks are computed with the daily closing price of the stocks then the Euclidean distances of the stocks are determined. Based on the distances, the minimum spanning tree of the stocks are constructed using the Prim algorithm. The results show the classification of the stocks. Then, the minimum spanning tree is refined to construct the dynamic minimum spanning tree sequences by computing with different steps and intervals. We find that the indicator "length/number" does not change with changing the steps and intervals. We propose a method to construct the synthetic tree and constructs the synthetic tree of the SSE50Index constituent stocks. The clustering results show that the proposed synthetic tree method is better than the minimum spanning tree constructed by Mantegna and Onnela.