| Products are constantly evolving in the market to better satisfy changing customers’ requirements.Acquiring and understanding the product preferences of customers in the market is the key to the successful new product developments.As one of the sources of understanding customer preferences,product sales data can provide a basis for the design and development of new products.Compared with traditional customer information collection methods such as interviews and questionnaires,sales data reflects customer preferences more objectively and comprehensively.Regarding product specifications,quantitative and qualitative analysis of customers preference information can facilitate the classification of customer needs and avoiding the errors caused by subjective factors.Advanced data mining technology can be used for mining potential customers preference information for supporting product design decisions.Based on the assumption that customer preferences are embedded in big sales data,this paper proposes a method of quantifying and mining customer preference information by establishing relationship between sales number and product specification combinations.On this basis,cluster analysis is carried out considering customer preference information.Then,design decision-making suggestions of modular design are provided through information entropy analysis of each cluster.Main contents of this research are as follows:1)At the level of product specifications,the information entropy formula is used to quantify customer preference information,including the information amount of single product specification and the information amount of product specification combinations.On this basis,the information content satisfying the constraint is defined according to the constraint of the product specification.2)Product sales data are collected and preprocessed.Those collected data are analyzed for the establishment of relationship model between product sale number and product specification combinations.The relationship model is used for the prediction of customers preference information.3)Based on the information of customers preference on product specifications,clustering is carried out for classifying the market into difference segments.For the minimal information content,product design suggestions are provided for better satisfying segmented market preference.4)Electric bicycle is taken as an example to illustrate effectiveness of the proposed method.Finally,further research directions are provided. |
