The invention discloses a garment matching recommendation method based on the integration of face attribute analysis. According to the invention, the shopping records and the browsing records of a user are comprehensively analyzed. Meanwhile, pictures which are shared by the user on a social network, and the face attribute characteristics of the user are also comprehensively analyzed. After that,a whole set of garment matching is recommended to the user. Based on a parameterization model obtained through learning, the collocation degree between different garment products, and the matching degree between garments and face attributes are speculated. In this way, the matching of garments, high in fitness for the user, is recommended for the user. According to the invention, the tensor decomposition method is adopted for the interaction between users and commodities, between face attribute characteristics and commodities, and between different commodities. Moreover, the gradient decreasing method is adopted for solving the problem of the multi-modal characteristics of fashion commodities and learning a nonlinear function to map characteristic vectors from the feature space to the potential space.