Value Analysis of User Consumption Behavior Based on BiT

With the continuous development of big data and artificial intelligence, companies in various fields continue to develop new user consumption behavior value analysis algorithms to improve the possibility of obtaining customers. In terms of data preprocessing, according to the logical method of data and sample screening, a preprocessing of the information was made to realize the unification of data for subsequent applications, Liu Junyin visualizes and analyzes a massive number of user data and then fits the data to obtain the statistical distribution law of user characteristics. Considering the behavior variables of each group of users, Huang Xinyu constructs the fully connected neural network binary prediction model and PCA fully connected neural network binary prediction model. In the model improvement stage, Zhou Xiaoxiao creatively proposed a bit based hidden potential patio time feature extraction model. The results of our tests led us to the conclusion that the bit-based latent spatio-temporal feature extraction model has the best learning behavior.