The Research and Analysis in Decision Tree Algorithm Based on C4.5 Algorithm

On the base of introducing the Decision Tree, this article lists two main algorithms of Decision Tree. After describing the principle of briefly, it reveals the essential idea and the computing process of C4.5 Algorithm in detail. As constructing a Decision Tree on a group of instance data with the two algorithms, we get the principle of classification about the data. By comparing the two algorithms during the process of constructing the tree, we analysis the advantages and disadvantages of them finally. With the conclusion of the applied scope of each algorithm, it provides a theoretical basis for choosing an appropriate method in practical application.

I. Introduction

Classification is the process of mapping data items into a learning function in a previously defined class. There are two types of modeling of classification problems: the statistical attribute modeling of the summary set and the modeling based on logic. Using the statistical method, the classification problem can be modeled by summarizing the statistical properties of the sample set. The modeling based on logic use of Boolean values and comparative operations for true or false evaluation to express, rather than addition and multiplication of such arithmetic operations. Compared with other nonlogical methods, the classification of this modeling method is more accurate, and they contain a broader range of explanatory attributes. Decision Tree and Decision Rules are typical way to output classification methods in a logical model.