Inventory Forecast Of The Cross-border E-commerce Based On Demand Characteristics And Sequence Trends

In the era of big data,in order to improve the competitiveness of overseas markets,cross-border e-commerce enterprises have began to mine valuable business information from massive data through data mining algorithms.With the upgrading of consumption,cross-border e-commerce users put forward higher requirements for the service level of enterprises.As a key factor affecting inventory management,inventory forecasting has an important impact on improving customer service level.According to the analysis of the current stocking situation of cross-border e-commerce,this paper puts forward the inventory forecasting method of cross-border e-commercebased on demand characteristics and sequence trend,named C-A-XGBoost Model.First,the C-XGBoost prediction model based on demand characteristics is established.The demand feature is used to cluster the goods by Two-step Clustering algorithm,and the multi-classification stocking strategy is formulated by the ABC classification method.On this basis,the XGBoost is used to build the prediction model for each clustering data,so that the demand feature is introduced into the prediction model as the influencing factor of prediction.Secondly,the A-XGBoost prediction model based on sequence trend is established.The residual of the ARIMA model is predicted by rolling prediction through the XGBoost to correct the nonlinear part.Finally,aiming at the minimum sum of the square error,the prediction results of C-XGBoost and A-XGBooost are weighted and summed by the least squares method to obtain the final C-A-XGBoost prediction model.This paper compares the proposed C-A-XGBoost model with the ARIMA,XGBoost,C-XGBoost,A-XGBoost model.By using the data of Zhejiang Jollychic Cross-border e-Commerce platform,the prediction results of five models are analyzed and compared,and it is proved that the prediction performance of the C-A-XGBoost model is better than that of the other four models.In this paper,the proposed prediction model is applied to inventory forecasting of the cross-border e-commerce,which provides a scientific reference for helping crossborder e-commerce enterprises to achieve accurate stocking preparation,and improve customer satisfaction,thereby improving the competitiveness of overseas markets.