What three metrics would you look at to identify churn risk in a customer transactions dataset, and how would you structure your predictive model?
