Which sequence best represents an end-to-end data-science case study?

A. Choose colors, deploy, collect data, and then define the problem B. Train once, ignore errors, and report only training accuracy C. Create a graph, delete data, and estimate a population D. Define problem, acquire and explore data, prepare features, model, evaluate, communicate, and monitor