Skip to Main Content
In many applications, there are problems of small sample size and high dimensionality of data, for example, in traditional Chinese medicine syndrome classification of chronic gastritis. To attack these problems, this paper gives a method which combines data preprocessing and Bayesian networks. Firstly, data is divided into groups with hierarchical clustering. Then, principal component analysis technique is used to extract principal components of each group of the data. At last, the new principal components are used to train a Bayesian network classifier. Experiment results demonstrate that the method is feasible and effective.