An Improved ID3 Classification Algorithm Based On Correlation Function and Weighted Attribute

ID3 decision tree algorithm is a supervised learning model based on calculating the information gain to select the best splitting attribute, which is the main factor to construct a decision tree. The process of calculating gain takes into consideration only a current condition attribute and decision attribute, and the other condition attributes cannot be used to measure the attribute importance. Because of the above problem, an improved ID3 takes into consideration the connection between the current condition attribute and the other conditions attributes. An experiment is presented to compare our improved algorithm with the traditional ID3 algorithm. Experiment results show that our improved algorithm provides a decision tree with less number of leaves and higher predictive accuracy.