Bi-criteria Optimization Problem for Decision Trees: Cost Versus Cost

Abstract

In this chapter, we consider an algorithm which constructs the sets of Pareto optimal points for bi-criteria optimization problems for decision trees relative to two cost functions. We also show how the constructed set of Pareto optimal points can be transformed into the graphs of functions which describe the relationships between the considered cost functions. We discuss three applications of bi-criteria optimization for two cost functions: comparison of different greedy algorithms for construction of decision trees, analysis of trade-offs for decision trees for corner point detection, and study of derivation of decision rules from decision trees.