Forecasting Cross-border E-commerce Sales Based on Gaussian Process Regression

Abstract

In recent years, cross-border e-commerce is booming. In the cross-border e-commerce model B2C (Business to Customer), the transaction subject is the enterprise and the consumer, due to its characteristics of small batch, multi-batch, facing many customers and order dispersion, the enterprise needs to plan the future experience strategy based on the sales, and the prediction of sales becomes more and more important. In order to meet the enterprise's demand for sales forecasting, this paper establishes a Gaussian process regression model, and selects Squared Exponential Kernel, Rational Quadeatic Kernel and Matern Kernel to construct a combined kernel function for forecasting. Comparing the prediction results with other models, it is found that the Gaussian process regression model using the combined kernel function has better prediction effect than other models, and the prediction results can help cross-border e-commerce enterprises to make decisions.