In or out? How to detect and respond to outliers in your loss reserving data
Outliers present difficult questions to the reserving actuary. On the one hand, estimates may be highly sensitive to a single point. On the other, is it wise to discard observed data, however skewed? This talk will consider these questions and consider what established statistical methods can teach us. We will also see how various techniques perform when applied to simulated data.
Register
Our team will contact you with the event details.
In or out? How to detect and respond to outliers in your loss reserving data
Outliers present difficult questions to the reserving actuary. On the one hand, estimates may be highly sensitive to a single point. On the other, is it wise to discard observed data, however skewed? This talk will consider these questions and consider what established statistical methods can teach us. We will also see how various techniques perform when applied to simulated data.
Learn more about the speakers
Brian Fannin
Brian Fannin has been an actuary for over 25 years. The data lacks sufficient credibility for him to give a more precise estimate. Brian is an Associate of the CAS, a Certified Specialist in Predictive Analytics (CSPA), and he holds a Masters in Statistics degree. He has worked in a variety of roles in commercial insurance, both primary and excess, here in the US as well as Europe, London, and Asia. He is also the author of the book "R for Actuaries and Data Scientists with Applications for Insurance," published by Actex. He currently works for Akur8 supporting their reserving software.
