Closed Alectoria closed 4 years ago
Dear Victoria, note that qgam
estimates quantile regression models, that is models where there is no parametric assumption on the distribution of the dependent variable, y
. Hence, if you think that the response y
follows a beta, gamma, Gaussian or any other distribution you should fit a GAM model gam()
using mgcv
, not a quantile GAM. With qgam
you can model the distribution of y
without making any such assumption, that is without specifying any family. That is why the qgam
function does not have an argument called family
. But maybe I am misunderstanding your request?
This is a request for an enhancement: It would be great if qgam would have the same range of families as family.mgcv in gam, e.g. the beta regression family to fit true % data.