pyomo is a python optimization package with a straightforward API, and super cool.
The book makes the already-approachable API very approachable, and for me it served as a very helpful survey of optimization applications. Any machine learning geek—myself included—is familiar with the optimization of a loss function, but there are a lot of optimization applications I had little exposure to. E.g. disjunctive programming, dynamic optimizations of problems with derivative constraints, these were much less familiar to me. It’s always fun to explore how other disciplines solve problems, and as someone without any in-depth operations research background, I found the book very helpful in becoming more aware of other tools at my disposal.
Within a few hours, I had
pyomo and an open source optimizer installed on an EC2 box and running on a problem of interest.
(Under the hood,
pyomo passes problems to your choice of optimizer.)
Skimming this first and referencing it definitely made that easy.