Learning optimal decisions with classification trees

Minja Zorc (2009) Learning optimal decisions with classification trees. EngD thesis.

PDF Download (1642Kb)

Abstract

Learning classification and regression models is one of the most important subfields of machine learning. Classification and regression models are constructed from learning set and used to classify new examples. In practice, we would often need model, which proposes optimal decision, for instance the best therapy for a certain type of illness for a particular patient. The main aim of the diploma thesis is to adapt the algorithm for the construction of classification trees to construct trees that would not predict the outcome but rather the decision leading to the desired outcome. Besides that, we had to develop the methods for measuring the quality of such models, and use it to test them on synthetic and real-world data sets.

Actions (login required)