Real-Time Sales Forecasting Algorithm of Electronic Commerce Products in a Smart City Based on Weighted Naive Bayes

In the real-time optimization and prediction of online sales of electronic (e-)commerce products, because of the diversity of users, randomness of data, limited server capacity, and statistics, the similarity index of node attributes is constrained, so false user information cannot be accurately analyzed and the accuracy of user activity analysis is poor. This paper proposes a Real-Time Sales Forecasting Algorithm of Electronic Commerce Products Based on Weighted Naive Bayes, product sales of the electricity suppliers push information do participles, cleaning, such as pretreatment, according to the push of information architecture space model, after pretreatment in high-dimensional sparse will be the basis of space model electricity product sales push information into the automatic encoder. According to the learning and layer-by-layer abstract output of e-commerce, product sales push information feature vectors, select the feature words, calculate the intraclass dispersion and interclass clustering degree of the feature words, and complete the real-time prediction of online sales of e-commerce products based on weighted naive Bayes. The experimental results show that the proposed algorithm realizes the real-time prediction of online sales of e-commerce products and provides a scientific basis for the optimization prediction of users’ sales potential.