Forecast by Design: Decision-Oriented Time Series Analytics in the Age of AI

Preface 
How to Use This Book (For Instructors) 

Part I Thinking in Time: The Logic of Forecasting 

Chapter 1 The Logic of Time and Decision 
Why Forecasting Is a Design Problem, Not a Modeling Task 
Opening Story: The Oil Price Rollercoaster 
1.1 Why Forecasting Matters in Business and Policy 
1.2 Forecasting vs. Prediction: Thinking Through Time vs. Conditions 
1.3 The Four Analytical Pillars of Modern Forecasting 
1.4 Statistical Learning and Machine Learning: Two Ways to Build Trust 
1.5 Structured vs. Unstructured Data: Two Forms of Evidence 
1.6 AI and the Expansion of Forecasting Design 
1.7 Time as Data: Trend, Seasonality, and Noise 
1.8 Uncertainty and the Forecasting Decision Loop 
1.9 Forecasting-by-Design 
SkillBox 1: Seeing Demand Through Time 
LearningLab 1: Thinking Through Time 
DesignStudio 1: Designing a Forecast-Informed Decision Process 
Mini-Case 1: Government Revenue Forecasting Under Uncertainty 
Check Your Learning 1: Reasoning, Interpretation, and Design 

Chapter 2 Smoothing Signals for Decision Sensemaking 
Designing Fast, Interpretable Signals for Decisions
Opening Story: From Pandemic Curves to Pizza Crowds 
2.1 Why Smoothing Supports Decision Sensemaking 
2.2 How Smoothing Reveals Temporal Signals 
2.3 Flat and Fading Memory: Moving Average and Exponential Smoothing 
2.4 Choosing Responsiveness: Designing Memory for Decisions 
2.5 Managerial Trade-Offs: Accuracy, Agility, and Trust 
SkillBox 2: Smoothing for Direction, Not Prediction 
LearningLab 2: Interpreting Smoothed Signals with AI 
DesignStudio 2: Designing Directional Signals for Decisions 
Mini-Case 2: Acting on Imperfect Signals 
Check Your Learning 2: Smoothing for Direction, Not Prediction 

Part II Modeling the Future: Foundations & Production

Chapter 3 Seeing Structure in Time 
Decomposition as a Tool for Interpretation and Communication
Opening Story: Marriott’s Road Back to Full Occupancy 
3.1 Decomposition as a Way of Thinking About Time 
3.2 The Core Components of Temporal Structure 
3.3 Additive and Multiplicative Structure: How Components Combine 
3.4 Decomposition as a Communication Tool 
SkillBox 3: Seeing Structure in Real Data 
LearningLab 3: Reasoning About Structure with AI 
DesignStudio 3: Designing Structure into Decision Systems 
Mini-Case 3: Making Sense of Time in a New Context 
Check Your Learning 3: Seeing Structure in Time 

Chapter 4 Designing Visible Temporal Structure 
Forecasting with Trend and Seasonality as Explicit Components
Opening Story: Tokyo Electric Power and the Cost of Misreading Visible Structure 9
4.1 From Seeing Structure to Designing It 
4.2 What Explicit-Structure Forecasting Does 
4.3 STL as a Representative Explicit-Structure Method 
4.4 Additive and Multiplicative Thinking 
4.5 The Explicit-Structure Forecasting Workflow 
4.6 Business Interpretation: Why Visible Components Matter 
4.7 Contrast Learning: Smoothing Vs. Explicit Forecasting 
4.8 Boundary of the Chapter: What Visible Structure Cannot Resolve 
SkillBox 4: Forecasting with Visible Structure Using STL 
LearningLab 4: Reasoning with Explicit Structure 
DesignStudio 4: Designing a Planning Response Around Visible Structure 
Mini-Case 4: Choosing Whether Visible Structure Is Sufficient 
Check Your Learning 4: Visible Temporal Structure in Forecast Design 

Chapter 5 Designing Hidden Temporal Structure in Forecasting 
Dependence, Memory, and Implicit Time Behavior
Opening Story: Grid Memory at Tokyo Electric Power 
5.1 From Visible Structure to Hidden Structure 
5.2 Dependence, Memory, and Shock Behavior 
5.3 Why Stabilization Comes Before Dependence 
5.4 Implicit Structure Through ARIMA/SARIMA 
5.5 Comparing Explicit and Implicit Forecasting Design 
SkillBox 5: Forecasting with Hidden Structure (ARIMA/SARIMA) 
LearningLab 5: Using AI to Reason About Hidden Structure 
DesignStudio 5: Designing Forecast Governance When Structure Is Hidden 
Mini-Case 5: Choosing How to Handle Hidden Structure 
Check Your Learning 5: Invisible Temporal Structure in Forecast Design 

Chapter 6 Governing Forecasts Over Time 
Forward Validation, Residual Signals, and Forecast Evolution
Opening Story: Airbnb’s Balancing Act—When Forecasts Lose Their Footing 
6.1 Why Forecasts Fail Quietly 
6.2 Accuracy and Reliability Are Not the Same Question 
6.3 Residuals as Signals, Not Just Errors 
6.4 Forward Validation—Watching Forecast Behavior as Time Moves Forward 
6.5 Forecast Evolution—Maintain, Refit, or Rethink 
SkillBox 6: Governing Forecasts Through Residual Signals 
LearningLab 6: Using AI to Reason About Forecast Trust 
DesignStudio 6: Designing a Forecast Governance Process 
Mini-Case 6: When Forecast Trust Erodes 
Check Your Learning 6: Governing Forecasts Over Time 

Part III Forecasting in the Age of AI 

Chapter 7 Forecasting Systems in the Age of AI 
Extending Structure with Machine Learning and Generative Reasoning
Opening Story: The Moment Forecasting Changed at Netflix 
7.1 From Box–Jenkins to Forecasting Systems in the Age of AI 
7.2 Forecasting Systems in the Age of AI 
7.3 Hybrid and Generative Forecasting Systems 
7.4 Explainability, Bias, and Ethics in AI-Era Forecasting Systems 
SkillBox 7: One Dataset, Four Forecasting Systems 
LearningLab 7: Interpreting AI-Based Forecasting Systems 
DesignStudio 7: Designing a Governed Hybrid Forecasting System 
Mini-Case 7: When Forecasts Disagree: Designing Decisions Under AI Uncertainty 
Check Your Learning 7: Forecasting Systems in the Age of AI 

Part IV Workflow Integration 

Chapter 8 Designing Decisions with Forecasts 
An End-to-End Forecasting in Action, Accountability, and Uncertainty
(with a Capstone Project Assignment)
Opening Story: Forecasting Public Health Emergencies 
8.1 From Prediction to Perspective: Designing Forecasting Lenses 
8.2 NorthStar in Action—From Forecast Outputs to Decision Action 
8.3 From Signals to Action—Designing Thresholds 
8.4 From Models to Lenses—Interpreting Multiple Futures 
8.5 Designing for Decision Robustness 
8.6 From Forecasting Systems to Decision Systems 
SkillBox 8: Building a Decision-Ready Forecasting System 
LearningLab 8: Multi-Lens Forecasting and Decision Design 
DesignStudio 8: Designing a Forecast-Driven Decision System 
Mini-Case 8: Acting Under a Surge Scenario 
Check Your Learning 8: Designing Decisions Under Forecast Uncertainty 
Capstone Project: Forecast-by-Design 

Part V Governance & Judgement 

Chapter 9 Forecasting-by-Design 
From Analytical Capability to Institutional Discipline
Opening Story: Forecasting for Climate Policy 
9.1 From Forecasting to Decision Design 
9.2 The Four Analytical Pillars as Decision Anchors 
9.3 The Human Dimension in the Age of AI 
9.4 The Forecasting-by-Design Cycle 
SkillBox 9: Designing Decisions with Forecasts 
LearningLab 9: Forecast Governance and Institutional Decision Design 
DesignStudio 9: Designing a Forecast Governance System 
Mini-Case 9: When Forecasts Disagree 
Check Your Learning 9: Thinking in Time, Deciding by Design 

Part VI Resources and Appendices 
Appendix A Extensions of Explicit-Structure Forecasting 
Appendix B Extensions of Implicit-Structure Forecasting 
Appendix C Feature-Based Machine Learning Extensions 
Appendix D Deep Learning and Sequence Models (Orientation Only) 
Appendix E Choosing Forecasting Methods in Practice: A Decision Matrix 

Data Dictionaries 
Bibliography & Suggested Readings 
Glossary 
Index