A Data Preprocessing Method Applied to Cluster Analysis on Stock Data by Kmeans

Recent years, more and more data mining methods are involved in applications like stock price analysis or predication, etc. Kmeans is one commonly used algorithm in those applications. However, those applications only take the technical indices (indicators) as features of data, where may make some important information lost, like the cross of different curves formed by the same technical index with different parameters. In this paper, we propose one way to quantify the variation trend of different curves, which can make kmeans clustering algorithm more effective on stocks analysis.