Abstract: The present study correspond discrete design elements to fashion Kansei information, and taken men’s coat for example, collected consumers preference rates of all related design elements through questionnaires, and then designed the consumers Kansei intention matrix. After normalized the matrices data, used K-means clustering algorithm for data processing through WEKA, which was the professional data mining software. Finally, base on the output data to built four typical design models. The purpose of this study is process and analysis the choices of most consumers from lots of fashion Kansei information, and fitting out the styles which satisfy the emotional needs of most consumers and to obtain the representative "best design model". Through the method of the present study proposed, could reveal the preference of different types of consumer effectively, and screen out the important design elements of positive impact to consume behavior, and provide reference to fashion designers effectively.

Key words: men's coat, perceptual intention materx, emotional preference, WEKA, clustering model