Retail analytics: integrated forecasting and inventory management for perishable products in retailing

Table of contents

1 Introduction
1.1 Motivation
1.2 Problem Statement
1.3 Outline

2 Literature Review
2.1 Unobservable Lost Sales
2.2 Assortment Planning
2.3 Assortment Planning with Stockout-Based Substitution
2.4 Stockout-Based Substitution in a Fixed Assortment
2.5 Joint Pricing and Inventory Planning with Substitution
2.6 Behavioral Operations Management

3 Safety Stock Planning Under Causal Demand Forecasting
3.1 Introduction
3.2 Safety Stock Basics and Least Squares Estimation
3.2.1 The Single-Variable Case
3.2.2 The Multi-Variable Case
3.2.3 Violations of Ordinary Least Squares Assumptions
3.3 Data-Driven Linear Programming
3.3.1 The Cost Model
3.3.2 The Service Level Model
3.4 Numerical Examples
3.4.1 Sample Size Effects
3.4.2 Violations of OLS Assumptions
3.4.3 Real Data
3.5 Conclusions

4 The Data-Driven Newsvendor with Censored Demand Observations
4.1 Introduction
4.2 Related Work
4.3 Data-Driven Model with Unobservable Lost Sales Estimation
4.3.1 Cost Model
4.3.2 Benchmark Approaches
4.4 Numerical Examples
4.4.1 The Normal Distribution
4.4.2 The Negative Binomial Distribution
4.4.3 Sample Size Effects
4.4.4 Real Data
4.5 Conclusions

5 Data-Driven Order Policies with Censored Demand and Substitution in Retailing
5.1 Motivation
5.2 Related Work
5.3 Model
5.3.1 Data
5.3.2 Decisions
5.3.3 Objective Function
5.3.4 Known Demand with Stockout Observations of One Product
5.3.5 Censored Demand
5.4 Numerical Study and Empirical Analysis
5.4.1 Benchmark to Estimate Arrival Rates and Substitution Probabilities
5.4.2 Optimal Solution
5.4.3 Data Generation
5.5 Results
5.5.1 Known Demand with Stockout Observations of One Product
5.5.2 Censored Demand with Stockout Observations of One Product
5.5.3 Censored Demand with Stockout Observations of Both Products
5.5.4 Real Data
5.6 Conclusions

6 Empirical Newsvendor Decisions Under a Service Level Contract
6.1 Introduction
6.2 The Setting
6.2.1 Data Overview
6.3 Modeling Demand
6.4 Normative Decision Model
6.4.1 Product-Specific Service Level
6.5 Empirical Analysis
6.5.1 Expected Profit Maximization
6.5.2 Alternative Decision Models
6.5.3 Comparison of Alternative Decision Models with the Empirical Retailer
6.6 Additional Behavioral Aspects of Decision Making
6.6.1 Anchoring and Adjustment
6.6.2 Minimizing Ex-Post Inventory Error
6.6.3 Order Adaptation and Demand Chasing
6.7 Value of Product Characteristics: Managerial Insights
6.8 Conclusions

7 Conclusions
7.1 Summary
7.2 Limitations and Future Research Directions

This book addresses the challenging task of demand forecasting and inventory management in retailing. It analyzes how information from point-of-sale scanner systems can be used to improve inventory decisions, and develops a data-driven approach that integrates demand forecasting and inventory management for perishable products, while taking unobservable lost sales and substitution into account in out-of-stock situations. Using linear programming, a new inventory function that reflects the causal relationship between demand and external factors such as price and weather is proposed. The book subsequently demonstrates the benefits of this new approach in numerical studies that utilize real data collected at a large European retail chain. Furthermore, the book derives an optimal inventory policy for a multi-product setting in which the decision-maker faces an aggregated service level target, and analyzes whether the decision-maker is subject to behavioral biases based on real data for bakery products.

https://www.springer.com/gp/book/9783319133041