ENH: GEE with zero-inflated count models, generalized Poisson

[josef-pkt]

extensions to zero-inflated count models

Sarvi, Fatemeh, Abbas Moghimbeigi, and Hossein Mahjub. 2019. “GEE-Based Zero-Inflated Generalized Poisson Model for Clustered over or under-Dispersed Count Data.” Journal of Statistical Computation and Simulation 89 (14): 2711–32. https://doi.org/10.1080/00949655.2019.1632857.

Can we extend the GEE approach to general distributions, i.e. outside GLM families or LEF?

Because it's outside the LEF, estimating the mean function will not be orthogonal to other distributional assumptions, e.g. cluster correlation.

Currently we don't have any setup to handle cases like this.

[josef-pkt]

looks like a subset of issue #3900