Integrating Price Elasticity and Reinforcement Learning: A Data-Driven Framework for Strategic E-commerce Pricing

The retail business is an important aspect towards growth and general development. The growing competition among retailers makes customer acquisition and retention strategies increasingly significant. These strategies are based on proper pricing of products which has a direct impact on customer loyalty and revenue. The research paper examines reinforcement learning (RL) methods to dynamic e-commerce pricing based on the publicly available Retail Price Optimization dataset. Two RL-based pricing agents were Deep Q-Network (DQN) agent to price real-time dynamically and Monte Carlo Tree Search (MCTS) agent to price sequentially. Different metrics such as MSE, RMSE, MAE, MAPE, and R2were utilized for the evaluation of model performance and the reward analysis was applied to measure the consistency and stability of the model. The results showed that the DQN agent has a higher predictive accuracy and more stable pricing behavior as indicated by an R-squared value of 98.74, a minimum reward variance of 8.24, and an RMSE of 8.24. Ablation research integrating price elasticity information and RL agents also enhances accuracy and minimizes price deviation. The combination of reinforcement learning with elasticity-informed strategies is indeed beneficial, as it is proven through the fact that ensemble and hybrid reinforcement learning approaches always surpass single-agent techniques in terms of predictive accuracy and price variance, respectively. It has been observed that the flexibility of RL-based agents, particularly DQN, to adapt to dynamic conditions is at the same level and in many instances superior to traditional methods such as XGBoost, multiple linear regression, and decision trees under similar experimental conditions. The e-commerce pricing that gives practical strategies grounded on elasticity, profitability optimization, as well as the ability to make real-time, competitive pricing decisions based on the product categories.