Analytical Skills for AI and Data Science by Daniel Vaughan -- Read -- Imperial Library of Trantor

Index

Preface

Why Analytical Skills for AI? Use Case-Driven Approach What This Book Isn’t Who This Book Is For What’s Needed Conventions Used in This Book Using Code Examples O’Reilly Online Learning How to Contact Us Acknowledgments

1. Analytical Thinking and the AI-Driven Enterprise

What Is AI? Why Current AI Won’t Deliver on Its Promises How Did We Get Here?

The Data Revolution

The three Vs Data maturity models

Descriptive stage Predictive stage Prescriptive stage

A Tale of Unrealized Expectations Analytical Skills for the Modern AI-Driven Enterprise Key Takeways Further Reading

2. Intro to Analytical Thinking

Descriptive, Predictive, and Prescriptive Questions

When Predictive Analysis Is Powerful: The Case of Cancer Detection Descriptive Analysis: The Case of Customer Churn

Describing churn Predicting churn Prescribing courses of action to reduce churn

Business Questions and KPIs

KPIs to Measure the Success of a Loyalty Program

An Anatomy of a Decision: A Simple Decomposition

An Example: Why Did You Buy This Book?

A Primer on Causation

Defining Correlation and Causation Some Difficulties in Estimating Causal Effects

Problem 1: We can’t observe counterfactuals Problem 2: Heterogeneity Problem 3: Confounders Problem 4: Selection effects A/B testing

Uncertainty

Uncertainty from Simplification Uncertainty from Heterogeneity Uncertainty from Social Interactions Uncertainty from Ignorance

Key Takeaways Further Reading

3. Learning to Ask Good Business Questions

From Business Objectives to Business Questions Descriptive, Predictive, and Prescriptive Questions Always Start with the Business Question and Work Backward Further Deconstructing the Business Questions

Example with a Two-Sided Platform

Learning to Ask Business Questions: Examples from Common Use Cases

Lowering Churn

Defining the business question Descriptive questions Predictive questions Prescriptive questions

Cross-Selling: Next-Best Offer

Defining the business question Descriptive questions Predictive questions Prescriptive questions

CAPEX Optimization Store Locations Who Should I Hire? Delinquency Rates Stock or Inventory Optimization Store Staffing

Key Takeaways Further Reading

4. Actions, Levers, and Decisions

Understanding What Is Actionable Physical Levers Human Levers

Why Do We Behave the Way We Do? Levers from Restrictions

Time restrictions

Levers That Affect Our Preferences

Genetics Individual and social learning Social reasons: strategic effects Social reasons: conformity and peer effects Framing effects Loss aversion

Levers That Change Your Expectations

The availability and representativeness heuristics

Revisiting Our Use Cases

Customer Churn Cross-Selling Capital Expenditure (CAPEX) Optimization Store Locations Who Should I Hire? Delinquency Rates Stock Optimization Store Staffing

Key Takeaways Further Reading

5. From Actions to Consequences: Learning How to Simplify

Why Do We Need to Simplify?

First- and Second-Order Effects

Exercising Our Analytical Muscle: Welcome Fermi

How Many Tennis Balls Fit the Floor of This Rectangular Room? How Much Would You Charge to Clean Every Window in Mexico City? Fermi Problems to Make Preliminary Business Cases

Paying our customers for their contact info Excessive contact attempts increase the probability of churn Should you accept the offer from that startup?

Revisiting the Examples from Chapter 3

Customer Churn Cross-Selling CAPEX Optimization

Price effect Quantity effect

Store Locations Delinquency Rates Stock Optimization Store Staffing

Key Takeaways Further Reading

6. Uncertainty

Where Does Uncertainty Come From? Quantifying Uncertainty

Expected Values

Bidding for a highway construction contract Interpreting expected values

Making Decisions Without Uncertainty Making Simple Decisions Under Uncertainty Decisions Under Uncertainty

Is This the Best We Can Do? But This Is a Frequentist Argument

Normative and Descriptive Theories of Decision-Making Some Paradoxes in Decision-Making Under Uncertainty

The St. Petersburg Paradox Risk Aversion

Putting it All into Practice

Estimating the Probabilities

Estimating unconditional probabilities Estimating conditional probabilities A/B testing Bandit problems

Estimating Expected Values Frequentist and Bayesian Methods

Revisiting Our Use Cases

Customer Churn Cross-Selling CAPEX Optimization Store Locations Who to Hire Delinquency Rates Stock Optimization

Key Takeaways Further Reading

7. Optimization

What Is Optimization?

Numerical Optimization Is Hard Optimization Is Not New in Business Settings Price and Revenue Optimization

Optimization Without Uncertainty

Customer Churn Cross-Selling CAPEX Investment Optimal Staffing Optimal Store Locations

Optimization with Uncertainty

Customer Churn Cross-Selling Optimal Staffing Tricks for Solving Optimization Problems Under Uncertainty

Key Takeaways Further Reading

8. Wrapping Up

Analytical Skills

Asking Prescriptive Questions Understanding Causality

Thinking outside the box

Simplify Embracing Uncertainty

Do-nothing approach The data-driven approach The model-driven approach

Tackling Optimization

Understanding the objective function Dealing with local optima Sensitivity to initial guesses Scaling and production issues

The AI-Driven Enterprise of the Future

Back to AI

Learning how to make decisions Some problems with this approach to automatic decision-making Ethics

Some Final Thoughts

A. A Brief Introduction to Machine Learning

What Is Machine Learning? A Taxonomy of ML Models

Supervised Learning Unsupervised Learning Semisupervised Learning

Regression and Classification Making Predictions

Caveats to the Plug-in Approach Where Do These Functions Come From? Making Good Predictions

From Linear Regression to Deep Learning

Linear Regression

Controlling for other variables Overfitting

Neural Networks

Activation functions: adding some extra nonlinearity The success of deep learning

A Primer on A/B Testing

A/B testing in practice Understanding power and size calculations False positives and false negatives