Prediction Models of Polka Dot's Preference That Incorporate Consumer Attribute as a Prior Knowledge

Abstract

In this study, the construction technique of conjoint measurement model using an image attribute and their applications are described to predict consumer's preference.
To incorporate consumer's information into the prediction model as prior knowledge, the consumer attribute was calculated using the multivariate analysis of variance. Constructive elements of handle and those combinations were used as the product attribute of the prediction model, and they were defined as image attributes. As the applications, one-piece dress and necktie of monochrome polka dot handle were chosen. Finally, the selected conjoint measurement models were eight kinds of one-piece dress model and one kind of necktie model. Main results obtained are as follows:
(1) As a result of multivariate analysis of variance, four effects were extracted, i.e., age, purchase area, sex and the interaction between purchase area and sex.
(2) An appropriate conjoint measurement model was composed of two kinds of image attribute and those four levels.
(3) To examine the utility of the selected prediction models, those were evaluated with an image screen of the polka dot handle.
As a result, the agreement rate of the one-piece dress was high.
Therefore, we have considered that the prediction of handle's preference by conjoint measurement model, which incorporates the prior knowledge of consumers, is useful for merchandising of apparel products.