Previous studies have shown that multi-category product variety is one of the key variables to attract consumers to the store or online retail site to buy, and the presentation of these categories has a significant impact on the consumer selection process and final purchase. In practice, multiple large e-commerce platforms (For example, Jingdong, Taobao and Suning E-Buy) also introduce these two attribute dimensions as the basic way of attribute classification. Mixed attributes usually appear in the form of "comprehensive (comprehensive ranking combined with multiple single attributes)", while single attributes usually include (sales volume, price, number of comments or reputation) and other elements. These two attribute modes constitute the product screening and classification of consumers on e-commerce platform. Basic rules, and become a necessary tool for consumers to screen products. Therefore, for the e-commerce platform and its merchants, clarifying the consumer's behavior feedback on the product selection set of different attribute classification modes is helpful for the e-commerce platform to optimize and perfect the existing attribute classification system, so as to provide better product screening service for consumers. At the same time, the study of unplanned purchases on e-commerce platforms is not only conducive to the development of e-commerce platforms and sales growth, but also helps to promote the progress and development of the entire e-commerce market. In view of this, this study focuses on the impact of the classification presentation of online product choice set (based on single attribute classification vs. based on mixed attribute classification) on consumers 'online unplanned purchase behavior and its internal mechanism. This paper focuses on the relationship between the attribute classification of different product choice sets and consumers 'online unplanned consumption behavior, and explores the impact of different attribute classification of product choice sets (mixed attribute vs. single attribute) on consumers' online unplanned consumption behavior. The contents of each study are as follows: Study 1 uses real e-commerce platform sales data to preliminarily verify that the product selection set with mixed attributes increases consumers 'unplanned purchases compared with single attributes. Study 2 examined the mediate role of product perception differences in the relationship between product attribute classification and unplanned purchase behavior. In study 3, the moderator effect of consumer preference clarity was verified by manipulating the subjects 'preference clarity in laboratory experiments. Finally, Study 4 found through laboratory experiments that when consumers have a low budget focus, they are less likely to make unplanned purchases. In the first part of the study, Study 1 proves that the product choice set presented by mixed attribute classification can increase the unplanned purchase behavior of consumers more than the single attribute classification. Study 1 first used real consumer behavior data, and labeled it as unplanned purchase by observing the consumer goods category that each consumer only once (and this category belongs to daily consumables, not durable goods), which helps us to further explore the internal mechanism conditions of this effect in the next experiment. In the second part of the study, study 2 revealed that the higher non-planned purchase behavior induced by the mixed attribute classification presentation is mediated by the higher product perception difference of the product selection set. At the same time, we further exclude emotion, product visualization ease, and product choice set size perception as alternative explanation mechanisms. Although the explanation mechanism of perceived disfluency of product presentation is not directly excluded, the mediate effect of product perception difference is still significant after controlling for this variable. In the following study, we further explore whether there are corresponding moderator or boundary conditions for this effect. In the third part of the study, Study 3 further verified the boundary condition of the existence of the main effect, that is, only when consumers have no clear preference, the mixed product attribute classification will enhance the unplanned purchase behavior of consumers. However, when consumers have high preference clarity, they will focus on the preferred products, so the product presentation of different classification attributes will not have a different impact on their unplanned purchase behavior.