Explore the intricacies of counterfactual evaluation in machine learning models with Michael Manapat from Stripe in this 39-minute conference talk. Delve into the challenges of charge outcomes prediction and model building, examining various approaches to improve accuracy. Learn about the fundamental problems faced in ML model evaluation, including recall training and the development of better analytical techniques. Gain insights into new model iterations and technical considerations, ultimately enhancing your understanding of ML model assessment and refinement in real-world applications.