C4.5 Algorithm Processing
The input of the algorithm which is produced by decision tree, is a group of examples with category tags, the output results of construction is a tree of two branches or more. The following algorithm is the main algorithm for constructing decision tree, that method uses top done recursive structure. After getting the full-grown decision tree, it will adopt a pruning strategy with decision tree for eliminating noise data and abnormal branching is caused by isolated point. The pruning of decision tree is proposed for the problem of the training data is to much to adopted. The method of pruning usually deletes the least reliable branch by statistical methods, to improve the speed of classification and recognition, and the capacity of accurate data classification. The essence is to eliminate noise and isolated points, and do loop to the ending of the training set. As shown in Figure 1.