A regression-based Kansei engineering system based on form feature lines for product form design

Yan Xiong,Yan Li,Peiyuan Pan,Yu Chen

doi:10.1177/1687814016656107

Yan Xiong, Yan Li + Show 2 more

Open Access

https://doi.org/10.1177/1687814016656107

Copy DOI

Abstract

When developing new products, it is important for a designer to understand users’ perceptions and develop product form with the corresponding perceptions. In order to establish the mapping between users’ perceptions and product design features effectively, in this study, we presented a regression-based Kansei engineering system based on form feature lines for product form design. First according to the characteristics of design concept representation, product form features–product form feature lines were defined. Second, Kansei words were chosen to describe image perceptions toward product samples. Then, multiple linear regression and support vector regression were used to construct the models, respectively, that predicted users’ image perceptions. Using mobile phones as experimental samples, Kansei prediction models were established based on the front view form feature lines of the samples. From the experimental results, these two predict models were of good adaptability. But in contrast to multiple linear regression, the predict performance of support vector regression model was better, and support vector regression is more suitable for form regression prediction. The results of the case showed that the proposed method provided an effective means for designers to manipulate product features as a whole, and it can optimize Kansei model and improve practical values.

Highlights

When developing new products, it is important to understand user needs, desires, and preferences toward products
Product form features were constructed with form feature lines (FFLs), products with similar image can be constructed by FFLs according to Kansei engineering (KE), users’ perceptions were influenced by the parameters of FFLs in a quantitative manner, the prediction models including multiple regression analysis (MLR) and support vector regression (SVR) were established to connect product features and users’ perceptions, and the validation of this method was verified
In MLR and SVR with radial basis functions (RBF) kernel model, the results of t-test clearly showed no significant differences in the experiment

Summary

Introduction

It is important to understand user needs, desires, and preferences toward products. A regression-based KE system based on form feature lines (FFLs) was put forward, FFLs which extracted from design representation were defined, and linear prediction model (MLR) and nonlinear prediction model (SVR) were used, respectively, to construct the connections between FFLs and users’ perceptions. According to the above discussion, a regression-based KE system based on FFLs for product form design was put forward The case of mobile phone design was used as an example to demonstrate the proposed method The criterion for this type of product is based less on functionality yet more on user subjective assessment.

Design parameters

Findings

Conclusion