Modeling the Severity of Random Events with the SAS/ETS® SEVERITY Procedure
Mahesh V. Joshi, SAS Institute 2010
The new SEVERITY procedure fits probability distributions for the severity (magnitude) of random events. Important examples include events with negative impact (such as the distribution of losses claimed under insurance policies, the magnitude of damages caused by natural disasters, and the severity of outbreaks of a disease) and events with positive impact (such as order sizes for products characterized by intermittent demand).
The SEVERITY procedure can use any of eight different parametric families of probability distributions. If these eight built-in distributions are not sufficient for your problem, then you can extend the SEVERITY procedure to support any parametric family by specifying its CDF and PDF functions with SAS® programming statements and the FCMP procedure.
The SEVERITY procedure can model the effects of exogenous variables on the severity of an event. It can also model censored or truncated data, which is an important feature, especially for insurance problems. Data on payments under insurance policies are subject to deductibles (which cause left truncation of the data) and are also subject to coverage limits (which produce right censoring).