Demand Prediction, Predictive Shipping, and Product Allocation for Large-Scale E-Commerce
Posted: 27 Nov 2018
Date Written: November 1, 2018
In this paper, we first extensively analyze the transactional level data made available by Alibaba and its logistics arm Cainiao, including detailed information on transaction orders, inventory and logistics for 7,103 different products and 130 warehouses. Based on the data analysis, we focus on demand prediction, data-driven shipping mechanism, and product allocation across warehouses: (i) We develop a multiple-product demand prediction system that identifies unique features presented in the data, which improves prediction accuracy significantly compared to using standard machine learning models. A new clustering-based regularization method is developed with the aid of representation learning to augment data and prevent overfitting. (ii) We propose and analyze in theory a novel shipping mechanism - Predictive Shipping, which utilizes demand prediction to arrange shipping before orders are placed. The observed vast vacancy in regional warehouses are utilized to support this mechanism. (iii) We formulate a large-scale product allocation problem across warehouses and study the cost sensitivity of a change in allocation distributions. Numerical experiments with both real data and synthetic data are conducted to demonstrate our findings.
Suggested Citation: Suggested Citation