|Saturday, February 17|
|CS18 Survival Analysis v. 'Survival' Analysis||
Sat, Feb 17, 9:15 AM - 10:45 AM
'How Long Would You Wait?' Using Time-to-Event (Survival) Analysis to Explore Waiting Times (303562)
*Ruth Hummel, SAS Institute
Keywords: time to event, survival, reliability, Cox proportional hazards, Kaplan-Meier, lognormal, Weibull, hazard, model comparison, LASSO, elastic net, censoring, truncation
How long do Seattle residents wait for 911 rescue to arrive? How long do Starbucks customers wait in line for coffee? What is the expected progression of a certain disease? When do people abandon a slow-to-load webpage? How long will my light bulb last? Do the time-to-event trajectories for these questions differ for distinct groups of people? One modeling tool for dealing with these sorts of time-to-event questions is the class of survival analysis models, also called time-to-event or reliability models. In this talk we will cover the basics of survival analysis, as well as the traditional modeling options (nonparametric, semi-parametric, fully parametric with a variety of potential underlying distributions), modern variable selection techniques, and model comparison. Attendees should leave with a better understanding of the function and application of many possible time-to-event analysis models, including understanding why and how to handle censoring and truncation and how to choose an appropriate model.