New Papers Published on Variance Website
Three new papers have been published on the Variance website.
Text Mining in Insurance: From Unstructured Data to Meaning
By Diego Zappa Mattia Borrelli Gian Paolo Clemente Nino Savelli
We use text mining to extract information from accidents reports about the impact of legal or illegal drugs taken by drivers at the time of crashes.
NCCI’s 2014 Excess Loss Factors
By Dan Corro Yen-Chieh Tseng
We describe the 2014 changes to NCCI’s excess loss factor methodology. Excess loss factors are used in NCCI’s retrospective rating plan, as well as in aggregate and class ratemaking.
Loss Reserving Using Estimation Methods Designed for Error Reduction
By Gary G. Venter
Fitting curves across model parameters increases parsimony and reduces prediction error. This works even better in a Bayesian framework. Related software goes beyond general linear restrictions, allowing more distributional choices.