Ashis SenGupta氏によるセミナー

The requirement of negative correlation or dependency gives rise to some interesting and challenging problems with both the construction of as well as the inference for such probability models. Probability models for such variables are constructed as with supports defined on the positive half or the entire real-line, which could possibly be members of the exponential family. Some dependency results are derived for these models. Inference procedures are also developed for these models. Challenges for extensions of such models to the multivariate case are discussed. The proposed models and methods are illustrated by several real-life examples. In particular, application to situations where such models are applied to Highly Accelerated Life Testing (HALT) and to a biological problem from US Army are discussed.