North America

Blue Room - Session 3

1:00 to 2:00 p.m. Tuesday, June 3, 2014

Survival Regression: From Biostats to Engineering

Reliability analysis of field data typically involves modeling the survival probability as a function of the vehicle's time-in-service and mileage. In certain cases, however, a need arises to additionally model the survival probability as a function of fixed and/or time-dependent covariates. Examples of the former may include "assembly plant," "production month," "days before sale," etc. The latter could be "cumulative exposure before and after a TSB," "seasonality," etc.

While survival regression models are quite popular in medical and pharmaceutical studies, their application in engineering data analysis is much less common. The purpose of the discussion is to show the practical benefits of using such models in automotive reliability data analysis.

Key Words: Failure Rate Estimation, Survival Regression, Warranty Claim Predictions, Warranty Cost

Vasiliy V. Krivtsov

Ford Motor Company

Dearborn, Michigan