Editing Decision tree learning (section)

==Extensions==

===Decision graphs===
In a decision tree, all paths from the root node to the leaf node proceed by way of conjunction, or ''AND''. In a decision graph, it is possible to use disjunctions (ORs) to join two more paths together using [[minimum message length]] (MML).<ref>{{cite web | url=http://citeseer.ist.psu.edu/oliver93decision.html | title=CiteSeerX}}</ref>  Decision graphs have been further extended to allow for previously unstated new attributes to be learnt dynamically and used at different places within the graph.<ref>[http://www.csse.monash.edu.au/~dld/Publications/2003/Tan+Dowe2003_MMLDecisionGraphs.pdf Tan & Dowe (2003)]</ref>  The more general coding scheme results in better predictive accuracy and log-loss probabilistic scoring.{{Citation needed|date=January 2012}}  In general, decision graphs infer models with fewer leaves than decision trees.

===Alternative search methods===
Evolutionary algorithms have been used to avoid local optimal decisions and search the decision tree space with little ''a priori'' bias.<ref>{{cite book |last1=Papagelis |first1=A. |last2=Kalles |first2=D. |year=2001 |chapter=Breeding Decision Trees Using Evolutionary Techniques |title=Proceedings of the Eighteenth International Conference on Machine Learning, June 28–July 1, 2001 |pages=393–400 |chapter-url=http://www.gatree.com/wordpress/wp-content/uploads/2010/04/BreedinDecisioTreeUsinEvo.pdf }}</ref><ref>{{cite journal |last1=Barros |first1=Rodrigo C. |last2=Basgalupp |first2=M. P. |last3=Carvalho |first3=A. C. P. L. F. |last4=Freitas |first4=Alex A. |year=2012 |doi=10.1109/TSMCC.2011.2157494 |title=A Survey of Evolutionary Algorithms for Decision-Tree Induction |journal=IEEE Transactions on Systems, Man, and Cybernetics |series=Part C: Applications and Reviews |volume=42 |issue=3 |pages=291–312 |citeseerx=10.1.1.308.9068 |s2cid=365692 }}</ref>

It is also possible for a tree to be sampled using [[Markov chain Monte Carlo|MCMC]].<ref>{{cite journal |last1=Chipman |first1=Hugh A. |first2=Edward I. |last2=George |first3=Robert E. |last3=McCulloch |title=Bayesian CART model search |journal=Journal of the American Statistical Association |volume=93 |issue=443 |year=1998 |pages=935–948 |doi=10.1080/01621459.1998.10473750 |citeseerx=10.1.1.211.5573 }}</ref>

The tree can be searched for in a bottom-up fashion.<ref>{{cite book |last1=Barros |first1=R. C. |last2=Cerri |first2=R. |last3=Jaskowiak |first3=P. A. |last4=Carvalho |first4=A. C. P. L. F. |doi=10.1109/ISDA.2011.6121697 |chapter=A bottom-up oblique decision tree induction algorithm |title=Proceedings of the 11th International Conference on Intelligent Systems Design and Applications (ISDA 2011) |pages=450–456 |year=2011 |isbn=978-1-4577-1676-8 |s2cid=15574923 }}</ref> Or several trees can be constructed parallelly to reduce the expected number of tests till classification.<ref name="Tris" />