Sales Prediction based on Machine Learning

With the increasing influence of the Internet on people's life, the development of e-commerce platforms is more rapid, with users and earnings of these platforms showing a growing trend. In recent years, the strong support of national policies has also provided a good environment for the development of the e-commerce industry. Under the impact of the epidemic this year, the role of the e-commerce industry in the development of the national economy has become more prominent. In such cases, the number and the competitiveness of e-commerce platforms and e-commerce enterprises are increasing. If a platform wants to maintain its advantage in the competition, it must be able to better meet the needs of users, and do a good job in all aspects of coordination and management. At this point, the accurate forecast of the sales volume of e-commerce platforms is particularly important. At present, there are many studies on e-commerce sales prediction, but we are still exploring the prediction model that can be better applied in different scenarios. In this paper, we try and evaluate two linear models, three machine learning models and two deep learning models, finding that machine learning and deep learning models have no advantage in improving the accuracy of sales forecast, but on a predictive basis, models perform better when they include information on calendar and price.