Key Factors' Clustering for Records with Mixed Data

A table usually contains numerical and symbolic data, so a method is proposed for clustering the records of the table. First, by using the improved function (TF-IWF) of the term frequency-inverse document frequency (TF-IDF), the symbols are transformed into the corresponding numerical values, and the remaining numerical values remain unchanged. On the basis of data transformation, the principal factor iteration method and maximum variance rotation method are used to discover the key factors that affect the clustering of the records. And the scores of the key factor are obtained by the least square method. Then k-means algorithm is used to cluster the scores, and finally the clustering results of the records are obtained. experimental results show that the evaluation indexes of the proposed method are better than those of other clustering methods, and the proposed method is helpful to dimensionality reduction and extract the latent important features of different clusters, so that each cluster can be labeled precisely.