Abstract: The prediction models of ecommerce website users’ behavioral intentions were established. The three aspects of users’emotional experiences (subjective emotional experience， eye movement characteristics and physiological responses) while browsing an ecommerce website the first time were measured， and their behavioral intentions (approach tendency) for the site thereafter were analyzed. Oneway analysis of variance was used to extract the indicators of emotional experiences affecting users’ behavioral intentions and the relationship models of emotional experiences and behavioral intentions were established based on the support vector machine (SVM). The results show that an integrated use of such indicators as subjective emotional evaluation， eye movement and physiological response can better predict users’ website approach tendency.

Key words: multimode measurement, ecommerce website, emotional experience, behavioral intention, support vector machine(SVM)