The greater the gain of the information, in general, indicates that the larger the "purity gain" resulting from the division of the dataset by the features used. Therefore, the information gain can be used to make the decision tree division property selection, in fact, is to select the information gain the largest property, ID3 algorithm is to use such information gain to divide the property.THE ID3 ALGORITHM IS BASED ON THE OCAM RAZOR PRINCIPLE, THAT IS, WITH AS LITTLE AS POSSIBLE TO DO MORE THINGS, IS A KIND OF DECISION TREE.