Before an AI system can do anything, a model is needed. In most cases, a model is equal to an environmental model. A powerful method of creating a model are decision trees.[1] Decision trees are explained in the literature for two applications. The first one is a replacement for a policy. That means, the decision tree results into the decision making process of the agent. It's equal to a behavior tree. The alternative is, to use decision trees for simulating a system. Then the Decision tree is equal to a physics engine.

The second understanding of a model isn't very often described in the literature. Most authors are focused on decision making with a direct policy which is encoded in the tree. But the mathematical background is the same. The idea is, that features are determined, for example the size of an egg, and the data are clustered around the features.[2] This allows to add new data in the existing group.


  1. Akshay Chavan: A Comprehensive Guide to Decision Tree Learning
  2. Gevrekçi, Yakut, and Cigdem Takma. "A Comparative study for egg production in layers by decision tree analysis." Pakistan Journal of Zoology 50.2 (2018).
Community content is available under CC-BY-SA unless otherwise noted.