Clustering-Based Data Preprocessing's Impact On Fuzzy Decision Tree Induction

Decision tree is the most widely used algorithm in commercial data mining tools. It's a method for approximating discrete-valued target functions, in which the learned function is represented by a decision tree. This model can denote the pattern of describing classification, and be used to accomplish prediction tasks. Decision tree is at present the main technique in the prediction and classification, especially in commercial fields, which has been successfully applied to a broad range of tasks from learning to diagnose medical case to learning to assess credit risk of loan applicants. In short, decision tree can help you transform data into knowledge.Introducing fuzzy mathematics makes decision tree technique be applied in broader technology applications and process more types of data. The fuzzy data processing methods, as an important component of fuzzy decision tree system, also have an important impact on the performance of fuzzy decision tree, which is also the focus of this study. The study founds that the membership function's location of the distribution of the fuzzy methods impact the effects of data fuzzification, thus affecting the implementation of fuzzy decision tree and its efficiency, accuracy and size. Kohonen clustering algorithm can be used to iterate to select the optimal centers of attributes with continuous value, and determine the location of membership function. Experimental studies have shown that the mechanism of the introduction of clustering to data fuzzy processing can adjust the transition regions, which make the coverage between fuzzy subsets no longer the same. This clustering algorithm can more reasonably describe the overlap relations between the fuzzy concepts. Compared with the other existing methods, this algorithm shows the achieved better accuracy of the fuzzy decision tree.