Study on Handling Range Inputs Methods on C4.5 Algorithm

How to successfully build a decision tree remains a focused topic in data mining. Hitherto many scholars have contributed a lot in the betterment of decision tree building algorithms. However, sometimes dataset may have range input attributes and present decision tree building methods, namely mean substitute, min-max substitute and mean-extent substitute, may not be suitable. This paper combines C4.5 and fuzzy mathematics to put forward a method structure in handling range inputs. The new method has important improvements on membership grade and entropy calculation method. Then a validation of the usefulness of the method is presented. The method is thought to be successfully applied to the investigation methodology, mainly in continuous data inputs with inexact data which consists of maximums and minimums.