Profiling a segment structure may help to make segments accessible and generally contribute to improving the understanding of the segments, thus helping to support management decisions. Logical type discriminant models yield results that are easy to understand; this makes them particularly useful for profiling. From this point of view, the induction of propositional rules yields better results when compared with hierarchical structures derived from decision tree algorithms. An application in the tourism market illustrates some advantages of the C5 algorithms for the induction of classification trees and sets of propositional rules for profiling a segment structure.

--

• Economia e Gestão - Ciências Sociais