Incremental Decision Tree Algorithm
- Most decision tree methods take a complete data set and build a tree using that data. This tree cannot be changed if new data is acquired later. Incremental decision trees are built using methods that allow an existing tree to be updated or revised using new, individual data instances. This is useful in several situations: a) the entire dataset is not available at the time the original tree is built, b) the original data set is too large to process, or c) the characteristics of the data change over time.
- (Utgoff, 1994) ⇒ Paul E. Utgoff. (1994). “An Improved Algorithm for Incremental Induction of Decision Trees.” In: Proceedings of the Eleventh International Conference on Machine Learning (ICML 1994).
- (Utgoff, 1987) ⇒ Paul E. Utgoff. (1987). “Id: an Incremental Id3." Technical Report. University of Massachusetts, Amherst, MA, USA.