Loading...

A Promotion Optimization Model in Retail Markets using Machine Learning Approach

Asadi, Ali | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57948 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Sedghi, Nafiseh
  7. Abstract:
  8. Determining a promotion planning is a critical decision for retail managers. This plan should decide on the amount and duration of promotions for each product in a way that maximizes profit compared to a non-promotion scenario. In this study, the promotion optimization problem in a retail environment is formulated as a non-linear integer programming problem. The objective function is to maximize profit from product sales during the sales period. The problem also includes several business-related constraints that limit the number of promotions. In this study, a reinforcement learning approach, specifically Deep Q-Network, has been used to solve the mathematical model. The implementation employs an epsilon-greedy policy for action selection and a prioritized experience replay buffer for storing experiences. Furthermore, to predict demand for profit calculation within the reward function of the DQN, a Long Short-Term Memory (LSTM) based demand forecasting model is used. The results, evaluated using the R-squared metric, indicate that the developed model can identify 74.34% of demand variability, making it suitable for use in the reinforcement learning model. The performance of the proposed DQN method was also tested using real data from an online retailer for low and high dimensional problem instances under various constraint scenarios with no constraints, strict constraints, moderate constraints and lenient constraints.. The results demonstrate that this approach can increase profits by 6.58% to 48.94% in low-dimensional cases and by 1.2% to 44.57% in high-dimensional cases compared to the baseline prices (no promotion)
  9. Keywords:
  10. Pricing ; Promotion Optimization ; Machine Learning ; Reinforcement Learning ; Demand Forecasting ; Long Short Term Memory (LSTM) ; Deep Q-Network (DQN)Algorithm ; Ret Ailers

 Digital Object List

 Bookmark

No TOC