Research on incremental decision tree algorithm

For data analysis of increase rapidly customer behavior, Web log analysis, network intrusion detection systems and other online classification system, how to quickly adapt to new samples is the key to ensure proper classification and sustainable operation. This paper presents a new adaptation data incremental decision tree algorithm, which combines RAINFOREST structure. It combines with the traditional SPRINT decision tree algorithm, and uses new samples quickly train a new decision tree based on the original decision tree. The improved algorithm deal with new samples at any time to produce a decision tree related, and the tree has been optimized with real-time.