Learning design preferences through design feature extraction and weighted ensemble

Abstract

Traditional custom manufacturing is made to order, usually on a small scale, to meet specific customer requirements, typically associated with higher costs because it involves producing goods in only small quantities. Mass production, on the other hand, costs less, but it does not reflect individual needs by producing uniform products. Because of the high cost of traditional custom manufacturing and the uniformity of mass production, both are not appropriate as a solution to the increasing demand for customized products. On the contrary, mass customization, a compound of customization and mass production, is a manufacturing strategy that allows manufacturers to produce customized products in large volumes while keeping costs low. This strategy combines the efficiency of mass production with the flexibility of custom manufacturing to meet the growing needs of the market for customized products. Design is a factor that must be considered for mass customization and plays an important role in consumer purchase decisions. As the need for understanding and predicting various preferences for each customer increases along with the importance of mass customization, predicting individual design preferences has become a critical factor in product development. However, current methods for predicting design preferences have some limitations. Product design involves a vast amount of high-dimensional information, and personal design preference is a complex and heterogeneous area of emotion unique to each individual. To address these challenges, we propose an approach that utilizes dimensionality reduction model to transform design samples into low-dimensional feature vectors, enabling us to extract the key representational features of each design. For preference prediction models using feature vectors, by referring to the design preference tendencies of others, we can predict the individual-level design preferences more accurately. Our proposed framework overcomes the limitations of traditional methods to determine design preferences, allowing us to accurately identify design features and predict individual preferences for specific products. Through this framework, we can improve the effectiveness of product development and create personalized product recommendations that cater to the unique needs of each consumer.