I am pleased to announce that my book
Sequential Decision Analytics and Modeling: Modeling with Python (SDAM)
is now available for free download from
https://tinyurl.com/sdamodeling/
The print version is also available from the webpage at a discounted price of $50 (use the special code on the webpage).
SDAM was written for my undergraduate course at Princeton University. It uses a teach-by-example style - with the exception of chapters 1 and 7, each chapter describes a different application using the same outline, making the book very easy to skim.
The book is designed to illustrate a fundamentally different way of approaching and solving sequential decision problems than has been used in the academic literature. It uses a “model first, then solve” style which is common in deterministic optimization, but which has not been used when making decisions over time under uncertainty. The goal of the book is to focus on teaching people how to think about sequential decision problems.
I recommend reading chapter 1 which presents the universal framework and sketches the four classes of policies for making decisions. Chapters 2-6 describe a series of examples that illustrates each of the four classes of policies, along with other modeling constructs. Chapter 7 then reviews some key concepts using the examples in the first six chapters. Chapters 8-14 present richer examples.
Most of the chapters are accompanied by Python modules to illustrate the model and policies. The Python modules are available on github here.
SDAM is designed to be used by itself. It is also a companion to my more advanced book:
Reinforcement Learning and Stochastic Optimization: A unified framework for sequential decisions