Model B

recall of 80% vs. 50%. It catches 80% of customers who will churn. 2. **Model A** — precision of 60% vs. 45%. Of the customers it flags, 60% actually churn (vs. 45% for Model B). 3. Suppose 100 customers, 20 will churn. Model A: catches 10 churners (recall 50%), flags 10/0.6 ≈ 17 total. Cost = 17 *