|Date:||Thursday, December 4, 2014|
In the reliability testing of many products, only the degradation of an associated quality characteristic is observed rather than the lifetimes of units directly. For analyzing data obtained from such degradation experiments, Wiener degradation and gamma degradation models have been discussed extensively in the literature. In this talk, I will first provide an introduction to the problem and then describe these two models. I will then introduce models with random effects that account for unit-to-unit variation and then discuss likelihood inference for this model and specifically describe the steps of the EM algorithm. I will use a laser data from the literature to illustrate all the inferential methods. Finally, I will discuss the optimal design of a future degradation experiment with budget constraints.