Title: Multivariate count distributions estimated by
iteratively reweighted Poisson regressions.
Supervisor: Thomas Yee
Prerequisites: STATS 310,330 and 782
Zhang et al. (2017), Journal of Computational and Graphical Statistics
26(1):1--13, proposed fitting several multivariate count distributions
by iteratively reweighted Poisson regressions (IRPR). This is because
the expected information matrices are expensive to compute. The VGAM
R package can, in theory, be adapted to perform IRPR because its main
algorithm is iteratively reweighted least squares. This project is to
accomplish this; some VGAM family functions need to be written by
adapting poissonff(). This project would suit a student with a good
mark in STATS 310 and 330, as well as knowing R well (STATS 782
because VGAM uses S4 object-oriented programming features).