skip to Main Content

An Introduction to the Analysis of Rare Events

Analyzing rare events like disease incidents, natural disasters, or component failures requires specialized statistical techniques since common methods like linear regression (PROC REG) are inappropriate. In this paper, we’ll first explain what it means to use a statistical model, then explain why the most common one (linear regression) is inappropriate for rare events. Then we’ll introduce the most basic statistical model for rare events: Poisson regression (using PROC GENMOD or PROC COUNTREG).

Published in 2011 and presented at the 2011 Western Users of SAS Software Conference and the 2011 MidWest SAS Users Group (MWSUG) Conference.

Click here for the PDF  
Back To Top