A Decision Tree Algorithm Based on Coordination Degree and Gini-Coefficient

The key to construct a decision tree is depending on how to select the split-attributes. This article presents an improved method which uses principal component analysis to consider comprehensive standard based on coordination degree and Gini-coefficient to overcome the flaw of ID3 algorithm that tended to choose attributes that had more values when chose split-attributes. The arithmetic enhances the effect of classification as experiment show. A final discussion argues for the utility of this algorithm about other aspect.