A clustering application method based on mix type variables in social system appraisement

Clustering is one of major function in data mining. Social system appraisement is faced with random variable and fuzzy variable, in order to solve the clustering of the mix type variables, this paper puts forward a clustering application method. The major thought of this method as follows: on the foundation of screening in advance for data source, classifies data variable according to whether belong to random variable or fuzzy variable. Random variable is determined by posterior probability distribution through Bayes study theory, and made it carry out clustering with the method of not random variable with posterior probability distribution as the weight of random variable. For fuzzy variable, the equivalence matrix is established according to the system clustering method of fuzzy relation, and converted it into Equivalence matrix to carry out clustering.