Define type I and type II errors for this classification problem. The insurance company cares more for missing the fraud cases than the other error possibility due to relative cost. Which model is better given this cost sensitivity?