E-commerce Sales Forecast Based on Ensemble Learning

Choujun Zhan, Jian‐Bin Li, Wei Jiang, Wei Sha, Yijing Guo

2020 · 15 citations · 12 references

Concepts

Abstract

E-commerce, driven by the technological advances of the semiconductor industry, becomes one of the fastest-growing industries and has become the largest sector of the electronics industry in recent years. Compared with the traditional offline business, E-commerce business eliminates geographical limitations by allowing consumers to purchase anywhere. Additionally, E-commerce only requires a web-based platform to start a business, resulting in a lower investment or business capital. Based on more than 75,000,000 transactions of one E-commerce company from January 2013 to June 2019, we analyze the economic impact of the holiday season, the relationship between price and sales, etc. Furthermore, we utilize Artificial Intelligence (AI) techniques to predict the sales of products. Here, assemble learning models, including Catboost, GBDT, LightGBM, and XGBoost, are adopted for regression prediction. The results show that Cat boost performs better than other methods.

References

12