I am working with counts data and I need the estimated coefficients for the mean to be non-negative for interpretation reasons.
So far, the only two packages support this are addreg and glmnet. However, addreg only suppports Negative Binomial 1 (NB1), but I want NB2 model.
As for the glmnet package, I have two problems about it: First, I keep encountering the error "algorithm did not convergeWarning". There are some threads about why this error would occur in binary data, but I have no clue as to why this would occur when I have count data or how to fix it. In addition, in glmnet, the size parameter is user-supplied rather than estimated. Since I don't have a good idea about what a correct or optimal size parameter is, I'd like it to be estimated.
Anyone has any idea about how I should do this? Thanks a lot!
