Dynamic Pricing Strategy for E-commerce Based on Reinforcement Learning

Abstract

Since 2025, with the rapid development and fierce competition of the e-commerce industry, the pricing strategy for e-commerce has already stepped into the stage of fine-tuning. Since the traditional fixed pricing strategy couldn’t adapt to market requirements, dynamic pricing has become the key to increasing profit. The research selects sales data from the Amazon platform in 2025 as an object of the study and uses current data processing and modelling approaches. By data preprocessing, data analysis, and eigenvalue mining, the research selects the Q-learning reinforcement learning algorithm to build a dynamic pricing model and conducts a control experiment on a traditional fixed pricing strategy. The results show that factors like price and sponsorship status have a significant influence on e-commerce sales volume, and there is a striking negative correlation between profit and sales volume. The total profit of the established dynamic pricing model, which is based on the Q-learning reinforcement learning algorithm, is 12.15% higher than the traditional fixed pricing strategy, which effectively increases profits. The research not only provides practical operation ideas and model support, but also proves the feasibility of the reinforcement learning algorithm in the e-commerce field.