Facing fierce competition in the marketplace, companies should try to determine the optimal settings of new products design. Human perception becomes the main objective in product development. This is related to consumer taste and preferences. In order to realize a well-designed product that appeals to consumers, the product not only should meet the physical requirements of consumers but also has to satisfy their affective needs. Thus, in a diverse and highly competitive market, successful products may not those with desirable product form. This study proposes Bayesian Rough Set model to deal some uncertainties and incomplete data in classification data. Bayesian Rough Set method has the strength to solve vagueness in human sensory data with the probabilistic approximation to identify the relationship rules between human perception and product exterior design. This is also used to handle heterogeneous population, contradictive between generalization versus customization, and uncertainty – inconsistency in market segmentation. At the final result, the proposed method get the better result than conventional method (Rough Set) as followed: The proposed method get accuracy improvement rather than conventional method (Rough Set), i.e. 12.07%, and confidence level as big as 10.61%.