0

I am working with high-dimensional count data, I have first fitted a Poisson regression in which the explanatory variables are log-transformed. I wanted to do the same with the Negative Binomial regression but I get this error:

model_nb<- glm.nb(m~log.transformed.counts, link="identity") 
Error: no valid set of coefficients has been found: please supply starting values
IRTFM
  • 258,963
  • 21
  • 364
  • 487
Mary
  • 3
  • 4
  • 1
    Identity-link count models are often numerically unstable/hard to fit. Can you explain why you're using an identity link? Can we have a [mcve] please? – Ben Bolker Mar 21 '19 at 16:51
  • https://stackoverflow.com/questions/31342637/error-please-supply-starting-values – Ben Bolker Mar 21 '19 at 16:56
  • if I put as link the log I get this error message: Error in glm.fitter(x = X, y = Y, w = w, etastart = eta, offset = offset, : NA/NaN/Inf in 'x' In addition: Warning message: step size truncated due to divergence" – Mary Mar 21 '19 at 17:08
  • OK, probably can't answer this without a [mcve] (see also https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example ) ... – Ben Bolker Mar 21 '19 at 19:37

0 Answers0