Integrating Stock Levels into Demand Forecasting and Discount Optimization for Retail | ||
| Industrial Management Journal | ||
| دوره 18، شماره 2، 2026، صفحه 183-207 اصل مقاله (1.45 M) | ||
| نوع مقاله: Original Research Article | ||
| شناسه دیجیتال (DOI): 10.22059/imj.2026.408958.1008279 | ||
| نویسندگان | ||
| Arezou Manosuri* 1؛ Mansour Momeni2؛ Mina Kazemi Miyangaskary3 | ||
| 1Ph.D. Candidate, Department of Technology and Innovation Management, Faculty of Industrial Management and Technology, College of Management, University of Tehran, Tehran, Iran. | ||
| 2Prof., Department of Technology and Innovation Management, Faculty of Industrial Management and Technology, College of Management, University of Tehran, Tehran, Iran. | ||
| 3Postdoctoral Researcher, Department of Management, University of Québec at Trois-Rivières, QC, Canada. | ||
| چکیده | ||
| Objective: Accurate sales forecasting is crucial for inventory management and supply chain planning in retail settings with high product variety, short life cycles, and volatile demand. Errors cause excess stock, stock-outs, lost sales, suboptimal promotions, higher costs, reduced customer satisfaction, and brand damage. Reliable forecasts drive replenishment, pricing, promotion design, inventory allocation, and profitability. As retailers embrace data-driven strategies, models must account for interactions among demand, pricing, and inventory. This study introduces a deep neural network framework that jointly tackles demand forecasting and discount optimization. Methodology: Unlike traditional approaches that ignore inventory limits, the model uses store-level stock availability as an input. This captures reality helping separate true demand from inventory-shortage effects. Experiments show this inventory-aware method substantially outperforms baselines without stock data on standard accuracy metrics. The framework also features a heuristic ladder search algorithm for discount optimization. It uses deep learning forecasts to evaluate discrete discount options, balancing demand uplift, remaining inventory, and profit margins. This prevents excessive markdowns that hurt margins or weak discounts that leave excess stock. Results: Tested on real-world data across products and stores, the approach yields better forecasts and notable profit gains by aligning pricing with inventory and demand. Conclusion: Overall, integrating demand forecasting and discount optimization outperforms separate handling, delivering retailers a practical tool to enhance inventory efficiency, promotion effectiveness, and profitability. | ||
| کلیدواژهها | ||
| Machine learning؛ deep neural network؛ parameter tuning؛ stock level features؛ discount ladder | ||
| مراجع | ||
|
| ||
|
آمار تعداد مشاهده مقاله: 331 تعداد دریافت فایل اصل مقاله: 299 |
||