Kernel principal component clustering methodology for stock categorization

A prerequisite of constructing an ecien t portfolio is to select some high-quality securities. It is therefore necessary to divide stocks into dieren t categories in terms of nancial information of listed com- panies. In order to categorize dieren t stocks and reduce the complexity of categorization, a kernel principal component clustering approach integrating kernel principal component analysis and k-means clustering is proposed for stock categorization. In the proposed approach, the sample data is rst preprocessed, and then the data dimension is reduced by kernel principal component analysis. Subsequently, the reduced data is used for clustering analysis using k-means clustering method. In terms of the clustering results, dieren t stock categories are nally obtained. For verication purpose, 20 listed securities from Chinese stock markets are used for empirical analysis. Experimental results revealed that the proposed kernel principal component clustering approach can obtain better categorization results than traditional clustering approaches.