Learning algorithm of decision tree generation for continuous-valued attribute

In this paper, we presented a learning algorithm using the information entropy minimization and mutual information entropy to generate a decision tree. Most of the current decision trees learning algorithms attempt to select the previously selected attributes for branching in a data set of which the values of condition attributes are continuous. The repeated selection limits the accuracy of training and testing. We should consider not only the information entropy minimization between the condition attributes and decision attributes but also the mutual information between condition attributes. The mutual information can avoid the repeated selection. The test results show that this method can obtain good performance.