In 2020, actuaries have at their fingertips more data, more computing power, and more statistical methods than ever before. As a result, the key question is no longer “How can we build a model?” More often, the key question is “Which model should we build first?”

In fact, a common problem actuarial departments face is having too many predictive models, with many struggling to be implemented, or providing answers still in search of the right questions.

This presentation and discussion will focus on a rational approach to building the right model at the right time, and defining model objectives to help actuaries and data scientists build more meaningful models for application in the P&C insurance process.Topics covered will include:

• Sources of uncertainty in the P&C Insurance process that can be used to help define model objectives
• A framework for defining and prioritizing predictive models to be built
• Discussion of examples (good and bad) of deriving meaning from the model
• Helpful “tweaks” to common modeling projects to enhance impact