Improvement Of C4.5 Decision Tree Algorithm

Since the C4.5 decision tree algorithm is more time-consuming and its high time complexity when processing continuous data.We put forward an improvement of C4.5 Decision Tree Algorithm method with the cloud model and the factor space theory.We propose a total discretization method based on the cloud modle.Above all,we deal with the value of sample which is continuous and elect the fundamental attributes.Then we discrete the attributes with the normal cloud model.We do experiments to test and verify the results.The results show this method is valuable and meaningful.We put forward a new heuristic function—the degree of ability to distinguish to construct a new decision tree algorithm more simply based on the factor space theory.We write the program in Matlab and do test with the UCI data.Then we compare with the C4.5 decision tree algorithm.The experiment result shows that the improved method is better than the C4.5 decision tree algorithm whether in aspect of the size and the accuracy.