Section 3

ML Engineering in Practice

Section 3 takes a working model through the decisions that follow the algorithm, measured on the diabetes and breast-cancer data: regularization, tuning and validation, rare classes and messy labels, faithful explanations, and the packaging, monitoring and release of a production pipeline. Chapter 18 finds where engineered features give way to learned representations on Banking77, and Chapter 19 reviews the volume through a fraud-scoring design.