User-driven segmentation of design data
Editor: Anja Maier, Stanko Škec, Harrison Kim, Michael Kokkolaras, Josef Oehmen, Georges Fadel, Filippo Salustri, Mike Van der Loos
Author: Maynard, Alex; Burnap, Alexander; Papalambros, Panos
Institution: University of Michigan, United States of America
Section: Design Methods and Tools
Design data is used to inform decisions during the design process, and must often be segmented for tasks such as customer segmentation, design benchmarking, and market preference segmentation. Qualitative data segmentation methods are accurate but not scalable due to being human-intensive, while quantitative segmentation models are scalable but often inaccurate due to mathematical assumptions. We propose Pangaea as an approach of combining human intelligence and computational algorithms, using an interactive 2D interface for user-driven segmentation on the frontend with both n-dimensional clustering and 2-dimensional reduction algorithms on the backend. We conduct an experiment segmenting automobile exterior color preferences. Our results show that users are able to find consistent data segmentations both between algorithms and between users, suggesting Pangaea may be a promising approach for combining human intelligence with computational algorithms for design data segmentation.