Over the past 10 years there has been a significant evolution of predictive analytics in actuarial analyses. These analytics provide us the opportunity to work with larger and more robust data. In response, the external and internal data sources and tools available have increased dramatically. Actuaries search this data with the idea that it may contain a new piece of information that will allow us to better price the underlying risk. This search is often called dimension reduction. This session will discuss different predictive modeling strategies and tools that are used in dealing with this issue, including standard parametric techniques to reducing dimensionality; non-parametric techniques including decision tree tools; pattern recognition in residuals; and a description of ongoing research and development as we continue to search for that needle in the haystack.

Moderator/Panelist:
Serhat Guven, Senior Consultant, Towers Watson