Skip to main content
2024 RPM Seminar Main banner
Final Program


Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Presentation Icons
Livestreamed
Livestreamed
Not Recorded
Not Recorded
Back
Evaluation ViewAttendees 1

RPM Seminar

Modeling

Level 1: No prior knowledge of the subject (0-3 Years)

C:MM:L1

C:MM:L2

C:MM:L3

S:CT:L1

S:BT:L2

T:IC:L3

T:CR:L3

M-7 - Penalized Regression and Lasso Credibility

Tuesday, March 19, 2024
2:45 PM – 3:45 PM CST
CE: 1.2
Livestreamed
Penalized regression is a powerful tool to enhance traditional GLM modeling. This session will explore the nature of this enhancement both in theory and in practice. We will briefly review GLMs before discussing the benefits of penalized regression and its deep connections to credibility. Then, we will take this one step further and describe the application of penalized regression directly as a credibility procedure: lasso credibility. By following the evolution of modeling from GLMs to lasso credibility, attendees will be able to identify practical applications of these techniques in their own work. The code for many examples from this presentation will be available on the CAS GitHub.

Learning Objectives:

  • Compare and contrast the underlying assumptions and review process of GLMs, penalized regression, and lasso credibility models
  • Understand the minimum-bias tradeoff in GLM and penalized regression and its relevance to credibility and out-of-sample predictive accuracy
  • Generate potential use cases of lasso credibility beyond the normal scope of GLM and penalized regression