Quantification of Effects of Complex Product Attributes Using Adaptive Questionnaire and Preference LearningCurrent students and collaborators: Kang, N., Ren, Y., Feinberg, F. and Papalambros, P.Y.
We investigate an adaptive questionnaire and learning mechanism to quantify the effects of complex product attributes (e.g., car styling) on consumer choices. In the context of car design, our method learns a mapping from shape parameters to a styling score, and another mapping from all car attributes (including styling) to consumer preference. All learning is performed in real time so that the questionnaire can find a near-optimal question to maximize the information gain. From the collected survey results, we further discover the relationship between consumer preference groups (e.g., people who are sensitive to MPG or price) and preferred vehicle styles.