Product form feature selection methodology based on numerical definition–based design

Abstract

In product design, the product form has a significant effect on the affective response it induces in potential consumers and is also of crucial importance if the product is to achieve commercial success. Intuitively, it seems reasonable to speculate that a consumer’s affective response to a product is dominated by certain critical features of the product form. To extract the product’s specific form features critical to determining consumers’ affective responses, this study proposes a product form feature selection methodology based on a numerical definition–based approach and the consumers’ affective responses. In the proposed methodology, numerical definition–based approach is used to generate an explicit numerical definition of the product form design, and the corresponding consumers’ affective responses (described using single adjectives) are determined by means of a semantic differential experiment. Two consumers’ affective response prediction models are constructed using support vector regression and multiple linear regression techniques, respectively. Finally, two feature selection methods, namely, support vector regression with support vector machine–recursive feature elimination and multiple linear regression with the stepwise procedure, are used to identify the critical form features. The validity of the two feature selection methods is demonstrated using a knife design for illustration purposes. The results show that the proposed methodology provides product designers with a powerful tool for systematically extracting the critical form features and evaluating their respective effects on the affective responses.