Markov Chain Monte Carlo Exact inference for binomial regression models
Current methods for conducting exact inference for logistic regression are not capable of handling large data sets due to memory constraints caused by storing large networks. We provide and implement an algorithm which is capable of conducting (approximate) exact inference for large data sets.
Various application fields, such as genetic epidemiology, in which logistic regression models are fit to larger data sets that are sparse or unbalanced may benefit from this work. We illustrate our method by applying it to a diabetes data set which could not be analyzed using existing methods implemented in software packages such as LogExact or SAS. We include a listing of our code along with documented instructions and examples of all user methods. The code will be submitted to the Comprehensive R Archive Network as a freely available R package after further testing.
This type of interdisciplinary work is a hallmark of our program in Applied Statistics at Simon Fraser University. For more information, please contact David Zamar (firstname.lastname@example.org) or his supervisor Jinko Graham (email@example.com), Department of Statistics and Actuarial Science.