Multi-Stage Customer Preferences Modeling Using Data-Driven Network Analysis

Cui, Yaxin

doi:https://doi.org/10.21985/n2-mye9-bg06

Work

Multi-Stage Customer Preferences Modeling Using Data-Driven Network Analysis

Public

Download PDF

Download All Files (.zip)

This dissertation aims to develop innovative analytical methods that integrate engineering, marketing, and social science disciplines to incorporate heterogeneous consumer preferences into product design using network-based customer preference modeling. Both companies and designers frequently face difficulties in understanding and addressing customer preferences, which can result in product failure and loss of market share. To overcome the limitations of existing methodologies, this dissertation presents a novel approach emphasizing network-based methods for modeling and analyzing customer preferences in engineering design and market research. By representing relations between customers and products as intricate networks and utilizing data-driven network analysis, this approach facilitates a deeper understanding of customer preferences. Consequently, it enhances product design and marketing strategies by effectively employing network-based techniques for preference modeling. The proposed approach comprises several key methodological developments, all focused on the concept of network-based customer preference modeling. First, a weighted network modeling approach for product competition analysis that captures the competition strength is introduced. This approach utilizes weighted network modeling and predictive analytics to examine product competition. Importantly, by quantifying the link strength in the network, this method offers a detailed understanding of the competitive landscape, identifying factors contributing to product success or failure in the market. Second, a framework incorporating information retrieval and survey design is developed to investigate customers' two-stage decision-making processes and the influence of their social networks. This framework captures the intricacies of consumer preferences and decision-making, revealing how preferences differ in the consideration and choice stages, and how social influence affects these preferences. The effectiveness of the proposed approach is demonstrated through a case study on household vacuum cleaners, highlighting its ability to capture consumer preferences and guide product design decisions. Third, a network-based analysis of heterogeneous customer preference modeling with market segmentation is presented. The proposed techniques enable an examination of varied customer preferences, aiding businesses in creating products that appeal to distinct market segments. Understanding the hierarchical structure of preferences and the underlying decision-making processes enables companies to customize marketing strategies effectively, targeting specific customer groups. Lastly, graph neural network-based methods in Link Prediction are investigated, concentrating on unidimensional product competition networks and preliminary findings on bipartite customer consideration-then-choice networks. These methods exhibit the potential for predicting consumer preferences and choices, providing valuable insights for both product design and marketing strategies. By integrating these advanced machine learning techniques, the proposed approach demonstrates its capacity to reveal complex patterns in consumer preferences, leading to a more comprehensive understanding of customer behavior. The proposed methodology equips engineering designers with the methodology and tools to better understand and respond to customer preferences and market trends, leading to more effective product design and marketing strategies. The contributions of this research have significant implications for both academia and industry, particularly in improving the design and marketing of consumer products. By employing network-based customer preference modeling, this dissertation offers an innovative approach to comprehending the complex nature of consumer preferences and their influence on product success. By highlighting the significance of network-based customer preference modeling and demonstrating its effectiveness through various contributions, this dissertation lays a robust foundation for future work in this area. Expanding these analytical methods to other industries and exploring additional network-based techniques will further enhance our understanding of consumer preferences, driving the development of successful products that cater to diverse customer needs. As the field continues to evolve, the insights gained from this research will play a crucial role in shaping the future of product design and marketing strategies.

Creator