Demand Prediction in Retail: A Practical Guide to Leverage Data and Predictive Analytics

Description

From data collection to evaluation and visualization of prediction results, this book provides a comprehensive overview of the process of predicting demand for retailers. Each step is illustrated with the relevant code and implementation details to demystify how historical data can be leveraged to predict future demand. The tools and methods presented can be applied to most retail settings, both online and brick-and-mortar, such as fashion, electronics, groceries, and furniture. This book is intended to help students in business analytics and data scientists better master how to leverage data for predicting demand in retail applications. It can also be used as a guide for supply chain practitioners who are interested in predicting demand. It enables readers to understand how to leverage data to predict future demand, how to clean and pre-process the data to make it suitable for predictive analytics, what the common caveats are in terms of implementation and how to assess prediction accuracy. Preface Acknowledgments Contents About the Authors Chapter 1: Introduction 1.1 Motivation 1.2 Dataset 1.3 Objective and Scope 1.3.1 Training and Test Data 1.3.2 Prediction Accuracy Metrics 1.3.3 Application References Chapter 2: Data Pre-Processing and Modeling Factors 2.1 Dealing with Missing Data 2.2 Testing for Outliers 2.3 Accounting for Time Effects 2.4 Price and Lag-Prices 2.5 Featured on Main Page 2.6 Item Descriptive Features 2.7 Additional Features 2.8 Scaling 2.9 Sorting and Exporting the Dataset References Chapter 3: Common Demand Prediction Methods 3.1 Primer: Basic Linear Regression for One SKU 3.2 Structuring the Dataset 3.3 Centralized Approach 3.4 Decentralized Approach 3.5 Feature Selection and Regularization 3.5.1 Subset Selection 3.5.1.1 Presentation of Subset Selection 3.5.1.2 AIC and BIC 3.5.2 Lasso Regularization 3.5.2.1 Lasso for One SKU 3.5.2.2 Decentralized Lasso 3.5.3 Ridge Regularization 3.5.3.1 Ridge Model 3.5.3.2 Decentralized Ridge 3.5.4 Elastic Net Regularization 3.5.4.1 Elastic Net Model 3.5.4.2 Decentralized Elastic Net 3.6 Log Transformations 3.6.1 Log-Transformation on the Price Variable 3.6.2 Log-Transformation on the Target Variable 3.6.3 Transformations and Prediction Accuracy 3.7 Centralized Approach with SKU-Fixed Effects 3.8 Centralized Approach with Price-Fixed Effects 3.9 Centralized Approach with SKU-Price-Fixed Effects 3.10 Decentralized Approach with Aggregated Seasonality 3.11 Summary and Next Steps References Chapter 4: Tree-Based Methods 4.1 Decision Tree 4.1.1 Centralized Decision Tree 4.1.1.1 Selecting the Parameters 4.1.1.2 Focusing on the Best Model 4.1.1.3 Example of a Plotted Tree 4.1.2 Decentralized Decision Tree 4.1.2.1 Selecting the Parameters 4.1.2.2 Focusing on the Best Model 4.2 Random Forest 4.2.1 Centralized Random Forest 4.2.1.1 Selecting the Parameters 4.2.1.2 Focusing on the Best Model 4.2.2 Decentralized Random Forest 4.2.2.1 Selecting the Parameters 4.2.2.2 Focusing on the Best Model 4.3 Gradient-Boosted Tree 4.3.1 Centralized Gradient Boosted Tree 4.3.1.1 Selecting the Parameters 4.3.1.2 Focusing on the Best Model 4.3.2 Decentralized Gradient-Boosted Tree 4.3.2.1 Fine-tuning the Parameters 4.3.2.2 Focusing on the Best Model 4.4 Methods Comparison References Chapter 5: Clustering Techniques 5.1 K-means Clustering 5.1.1 Description of K-means Clustering 5.1.2 Clustering using Average Price and Weekly Sales 5.1.3 Adding Standard Deviations of the Clustering Features 5.2 DBSCAN Clustering 5.2.1 Description of DBSCAN Clustering 5.2.2 Clustering using Average Price and Weekly Sales 5.2.3 Adding the Standard Deviation of the Clustering Features References Chapter 6: Evaluation and Visualization 6.1 Summary of Results 6.2 Prediction vs. Actual 6.3 Varying the Split Ratio Chapter 7: More Advanced Methods 7.1 The Prophet Method 7.1.1 What is the Prophet Method? 7.1.1.1 How it Works 7.1.1.2 Illustration for One SKU 7.1.2 Forecasting with Prophet 7.1.2.1 Univariate Time-Series 7.1.2.2 Adding Features 7.2 Data Aggregation and Demand Prediction 7.2.1 Presentation of the DAC Method 7.2.2 Fine-Tuning the Hyperparameters 7.2.3 Interpretating the DAC Results References Chapter 8: Conclusion and Advanced Topics References