Decision Trees as a Complementary Tool for Conjoint Analysis:Application to Door Panels User Preferences

Conjoint Analysis (CA) is a technique that supports value analysis tools providing information about the relative demanded quality of each component. However, it is not easy to assess the partial effect of each component in different configurations. For this purpose, inductive learning techniques as decision trees (DT) could be used. In this work, inductive learning techniques (ID3) are compared with CA analysis as regards the perception of quality in door panels.

An experimental design with 16 door panels combining different possibilities for top roll, insert, upper and lower armrest, handle and map pocket was carried out. 63 subjects participated in the study. The results were analysed by means of CA techniques and ID3 inductive learning. Both techniques allowed determining the combination of elements that configure the best and worst levels of quality.