What makes working on new machine learning (ML) use cases so exciting, and at times so frustrating, is ML’s lack of hard and fast rules. A few aspects of the model development process can be codified; for example, data should always be separated into strictly disjoint training and test sets…