Real Time Prediction on Revisitation Behaviors of Short-Term Type Commodities

With the advent of Internet we have entered a new era, more and more enterprises take the network as the new profit growth point of the marketing channel. The target of network trading platform is to attract consumers firstly. In this paper, we first extract the five common features in customers’ visitation behaviors, i.e. commodity’s type popularity, commodity’s type revisitation ratio, user revisitation ratio, window repeat ratio and brand popularity. These features constitute the major factors that affect people’s short-term types of commodity revisitation behaviors. Based on features of revisitation behaviors, we propose method that can quickly predict consumers will have a types of commodity revisitation behavior. The method is based on Fuzzy Comprehensive Evaluation Method, namely FCEM-II. The experiment of our method adopts a large scale real and reliable data set, and the experimental results show that our proposed method are more accurate in the prediction tasks compared with the baselines.