A Dynamic Export Product Sales Forecasting Model Based on Controllable Relevance Big Data for Cross-Border E-Commerce

Abstract

To improve the prediction accuracy of export product sales, this paper constructs a dynamic export product sales prediction model based on controlled relevance big data for cross-border e-commerce to improve sales prediction’s scalability and dynamic evolution. Based on the traditional prediction model, a big data controllable clustering algorithm is used to divide the data into several macro-clusters by data dimensions to determine the number of class clusters and the location of centroids. The K-mean algorithm is used to estimate and categorize the indicators affecting the prediction online, to dig out the key factors affecting the prediction of export product sales, and to establish a dynamic prediction model. The analysis results show that the plausibility measure of the proposed model is 21.9, and the error coefficient is 5.1, which are the smaller values in the reference interval. The average prediction error ratio is 2.25%, the average confidence level is 93.05%, and the error efficiency between predicted sales and actual sales is only 0.98%. Thus, the prediction model proposed in this paper improves the prediction effect of export product sales and has high practical value.

Keywords

• big associative data
• dynamic predictive model
• controlled clustering algorithm
• k-means algorithm
• plausibility measure
• 62-07