Dynamic Pricing on E-Commerce Platform with Deep Reinforcement Learning: A Field Experiment

By Jiaxi Liu et al

Published on Dec. 10, 2019

Read the original document by opening this link in a new tab.

Table of Contents

ABSTRACT
Keywords
1 Literature Review
2 Methodology
2.1 Problem Formulation
2.2 Discrete Pricing Action Models
2.3 Continuous Pricing Action Model

Summary

In this paper, an end-to-end framework for dynamic pricing on E-commerce platform using deep reinforcement learning is presented. The framework models the dynamic pricing problem as a Markov Decision Process and makes significant contributions in extending the problem to continuous price sets, defining a new reward function, and addressing the cold-start problem. The approaches are evaluated through offline and online experiments, showing better performance than manual pricing strategies. The paper discusses related work in dynamic pricing, introduces the methodology including problem formulation and pricing action models, and presents results validating the approach. The conclusion highlights the effectiveness of the dynamic pricing framework on E-commerce platform.