The Research on Discretization Algorithm Based on Dynamic Hierarchical Clustering

The discretization problem of continuous attribute is an important topic of machine learning, and data preprocessing. The discretization algorithm based on dynamic hierarchical clustering proposed in the paper is an improvement of hierarchical clustering algorithm. First, the qualitative analysis of the algorithm is made. Second, clustering analysis of the random collection data according to the similarity degree is described and a kind of division on the domain is received. Indicated by the experiment, the discretization algorithm based on dynamic hierarchical clustering could be more reasonable and efficient to the division of continuous attribute.