A hierarchical choice modelling approach for incorporating customer preferences in vehicle package design. (30th March 2009)

• Record Type: Journal Article
• Authors: Kumar, Deepak; Hoyle, Christopher; Chen, Wei; Wang, Nanxin; Gomez-Levi, Gianna; Koppelman, Frank
• Is Part Of: International journal of product development. Volume 8:Number 3(2009)
• Page Start: 228
• Page End: 251
• Publication Date: 2009-03-30
• Subjects: demand modelling -- enterprise-driven design -- DCA -- discrete choice analysis -- pooled estimation -- model fusion -- nested logit -- customer ratings -- ordered logit -- model calibration -- complex design artefacts -- vehicle packaging design -- hierarchical modelling -- choice modelling -- customer preferences -- consumer choice behaviour -- engineering design attributes -- consumer selection attributes -- vehicle design; New products -- Periodicals; Technological innovations -- Periodicals; Research, Industrial -- Periodicals; 658.57505
• Languages: English
• ISSNs: 1477-9056

Abstract

The use of customer preference models to evaluate the economic impact of design changes and new product introductions has become prevalent in the literature. However, existing approaches do not sufficiently address the needs of complex design artefacts, which typically consist of many subsystems and components designed and manufactured with significant autonomy. Characteristics of complex systems, such as heterogeneity of consumer preferences throughout the system hierarchy, multiple sources of information and qualitative consumer-desired attributes, have not been adequately addressed. In this work, we propose a hierarchical choice modelling approach for complex systems to model customer preferences for attributes throughout the system hierarchy, and to subsequently predict consumer choice behaviours. A system of hierarchical models is used to link the design attributes used for engineering design to the attributes used by consumers to choose among competing products. The model framework utilises Discrete Choice Analysis at the top level to model customer choices and Ordered Logit regression at the lower levels to model ordinal survey responses as a function of product attributes. An approach for combining choice data from multiple sources based on the Nested Logit methodology is developed. The framework is demonstrated on the vehicle occupant package case study.