An intelligent fuzzy regression approach for affective product design that captures nonlinearity and fuzziness

Abstract

Affective product design aims at incorporating customers' affective needs into design variables of a new product so as to optimize customers' affective satisfaction. Faced with fierce competition in marketplaces, companies try to determine the settings in order to maximize customers' affective satisfaction with products. To achieve this, a set of customer survey data is required in order to develop a model which relates customers' affective responses to the design variables of a new product. Customer survey data is usually fuzzy since human feeling is usually fuzzy, and the relationship between customers' affective responses and design variables is usually nonlinear. However, previous research on modelling the relationship between affective response and design variables has not addressed the development of explicit models involving either nonlinearity or fuzziness. In this paper, an intelligent fuzzy regression approach is proposed to generate models which represent this nonlinear and fuzzy relationship between affective responses and design variables. In order to do this,  K.Y. Chan is the corresponding author. His email address is Kit.Chan@curtin.edu.au.we extend the existing work on fuzzy regression by first utilizing an evolutionary algorithm to construct branches of a tree representing structures of a model where the nonlinearity of the model can be addressed. The fuzzy regression algorithm is then used to determine the fuzzy coefficients of the model. The models thus developed are explicit, and consist of fuzzy, nonlinear terms which relate affective responses to design variables. A case study of affective product design of mobile phones is used to illustrate the proposed method. h refers to the degree to which the fuzzy linear model fits the data sets in developing the fuzzy model. Keywords: Fuzzy regression, evolutionary algorithm, affective product design

A framework to support requirements analysis in engineering design

“…Recently, there are works using QFD and modelling techniques to represent a relationship between customer requirements and design attributes of new products, e.g. (Amihud et al, 2007, Chan et al 2010. Nevertheless, the QFD method is best for collecting and refining functional requirements hence the "F" in its name.…”

A forward selection based fuzzy regression for new product development that correlates engineering characteristics with consumer preferences

“…b) Peters' fuzzy regression (P-FR) [30] is a new version of T-FR, where the estimated interval on the generated model is bounded by all samples and the generated model is effective on detecting presence of outliers. P-FR has been used to develop consumer preference models for mobile phone design [22]. c) Hybrid fuzzy least square regression (H-FLSR) [2] can be used to address the uncertainties caused by fuzzy and random natures of the samples.…”

“…Although the heuristic method namely genetic programming has been used to develop the models involving with significant regressors, the genetic programming requires a lot more computational model evaluations than the statistical regression methods do [22]. Therefore, in this paper, a novel fuzzy modelling method, namely forward selection based fuzzy regression (FS-FR) which incorporates the approaches of fuzzy least square regression [23] and forward selection regression [24], is proposed.…”

“…Also we have compared the genetic programming [22,27] and the proposed FS-FR based on a case study of new product development. The case study shows that better generalization capability can be obtained by the proposed FS-FR.…”

Random And Fuzzy Measure Of Unpredictable Construction Works

“…It was assumed that Z and U as unsharp sets, determined under conditions of fuzziness, may be in a certain relation with one another. As a result, the fuzzy relation was defined as follows [2]:…”