Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Study of Data-driven Methods for Inventory Optimization

Lee Yeung Ping, Patrick Wong, Tan Cheng Han

TL;DR

The study evaluates data-driven inventory methods—Time Series, Random Forest, and Deep Reinforcement Learning—across Lost Sales, Dual-Sourcing, and Multi-Echelon supermarket inventory models. By integrating Prophet-based forecasting, tree-based demand prediction, and deep Q-learning within these models, it offers a comparative view of forecast accuracy, adaptability, and cost impact, highlighting how each method handles perishability, lead times, and multi-echelon coordination. The findings show RF and Gradient Boosting delivering the strongest predictive accuracy in many settings, Time Series providing usable short-term forecasts, while DRL yields mixed results requiring careful reward design and richer feature engineering. Practically, the work informs practitioners about method suitability under specific inventory structures and motivates future hybrid approaches that combine forecasting strength with adaptive decision-making to improve service levels and cost efficiency in retail supply chains.

Abstract

This paper shows a comprehensive analysis of three algorithms (Time Series, Random Forest (RF) and Deep Reinforcement Learning) into three inventory models (the Lost Sales, Dual-Sourcing and Multi-Echelon Inventory Model). These methodologies are applied in the supermarket context. The main purpose is to analyse efficient methods for the data-driven. Their possibility, potential and current challenges are taken into consideration in this report. By comparing the results in each model, the effectiveness of each algorithm is evaluated based on several key performance indicators, including forecast accuracy, adaptability to market changes, and overall impact on inventory costs and customer satisfaction levels. The data visualization tools and statistical metrics are the indicators for the comparisons and show some obvious trends and patterns that can guide decision-making in inventory management. These tools enable managers to not only track the performance of different algorithms in real-time but also to drill down into specific data points to understand the underlying causes of inventory fluctuations. This level of detail is crucial for pinpointing inefficiencies and areas for improvement within the supply chain.

A Study of Data-driven Methods for Inventory Optimization

TL;DR

The study evaluates data-driven inventory methods—Time Series, Random Forest, and Deep Reinforcement Learning—across Lost Sales, Dual-Sourcing, and Multi-Echelon supermarket inventory models. By integrating Prophet-based forecasting, tree-based demand prediction, and deep Q-learning within these models, it offers a comparative view of forecast accuracy, adaptability, and cost impact, highlighting how each method handles perishability, lead times, and multi-echelon coordination. The findings show RF and Gradient Boosting delivering the strongest predictive accuracy in many settings, Time Series providing usable short-term forecasts, while DRL yields mixed results requiring careful reward design and richer feature engineering. Practically, the work informs practitioners about method suitability under specific inventory structures and motivates future hybrid approaches that combine forecasting strength with adaptive decision-making to improve service levels and cost efficiency in retail supply chains.

Abstract

This paper shows a comprehensive analysis of three algorithms (Time Series, Random Forest (RF) and Deep Reinforcement Learning) into three inventory models (the Lost Sales, Dual-Sourcing and Multi-Echelon Inventory Model). These methodologies are applied in the supermarket context. The main purpose is to analyse efficient methods for the data-driven. Their possibility, potential and current challenges are taken into consideration in this report. By comparing the results in each model, the effectiveness of each algorithm is evaluated based on several key performance indicators, including forecast accuracy, adaptability to market changes, and overall impact on inventory costs and customer satisfaction levels. The data visualization tools and statistical metrics are the indicators for the comparisons and show some obvious trends and patterns that can guide decision-making in inventory management. These tools enable managers to not only track the performance of different algorithms in real-time but also to drill down into specific data points to understand the underlying causes of inventory fluctuations. This level of detail is crucial for pinpointing inefficiencies and areas for improvement within the supply chain.
Paper Structure (47 sections, 24 figures, 11 tables)

This paper contains 47 sections, 24 figures, 11 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Tested Inventory Models
  4. Lost Sales Inventory Model
  5. Dual-Sourcing Inventory Model
  6. The Multi-Echelon Inventory Model
  7. Selected Algorithms
  8. Time Series Method
  9. Random Forests
  10. Deep Reinforcement Learning
  11. Methodologies in Lost Sales inventory model
  12. Time Series
  13. Code Design
  14. Results
  15. Random Forest
  16. ...and 32 more sections

Figures (24)

  • Figure : Figure 1
  • Figure : Figure 2,3,4
  • Figure : Figure 5
  • Figure : Figure 6,7
  • Figure : Figure 8
  • ...and 19 more figures