search

alexpkeil1 / qgcomp

QGcomp (quantile g-computation): estimating the effects of exposure mixtures. Works for continuous, binary, and right-censored survival outcomes. Flexible, unconstrained, fast and guided by modern causal inference principles
27 stars 6 forks source link

Qgcomp with families other than "binomial" and "guassian" #8

Open alexpkeil1 opened 4 years ago

alexpkeil1 commented 4 years ago

Other distributions are currently being tested. Currently, the "poisson" distribution will not produce errors when used with ggcomp.boot and qgcomp.noboot functions. There is some further testing that needs to be done to ensure that there are no remaining issues to resolve. Use with caution and, if possible, test that results make sense using simulated data!

alexpkeil1 commented 3 years ago

"poisson" has been used numerous times with no apparent issues. If using the bootstrap version of qgcomp, it's advisable to set MCsize to a value greater (say 10x) than the sample size when using the Poisson distribution.

alexpkeil1 commented 1 year ago

Late update: the Cox model version similarly performs well. If using the bootstrap version of qgcomp, it's advisable to set MCsize to a value greater (say 10-50x) than the sample size when using the Cox model version.

alexpkeil1 commented 1 year ago

Late update: Hurdle and zero inflated model versions similarly performs well. If using the bootstrap version of qgcomp, it's advisable to set MCsize to a value MUCH greater (say 100-1000x) than the sample size when using Hurdle and zero inflated models.