4

I have modeled claim frequency data using Poisson regression and claim amount using gamma. I have seen the Tweedie compound Poisson gamma distribution used to model aggregate claim data. I am quite new to R and I am trying best to model it using the Tweedie model in R. I have searched on the R codes and found the following ptweedie.series(q, power, mu, phi) and dtweedie.series(y, power, mu, phi). I have predictors in my data set. Can anyone suggest how to proceed from here?

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
user40494
  • 425
  • 5
  • 13
  • 1
    I believe this is primarily a programming question, and this better suited for Stack Overflow. – Fomite Feb 18 '14 at 15:24
  • This question is off-topic b/c it is about how to use R & not about the underlying statistical issues. However, it will be off-topic on SO as well, b/c it is not really a programming question & lacks a reproducible example. This question might be appropriate for the r-help-listserv. – gung - Reinstate Monica Feb 18 '14 at 15:54
  • 2
    This question appears to be off-topic because it is about how to use R to do something. – gung - Reinstate Monica Feb 18 '14 at 15:54
  • 2
    @gung: I (respectfully) up-voted to save the question for what, I think, is premature deletion: I think this O.P. is essentially searching for a statistical answer (difference between univariate and conditional models). I don't think the question will get *any* answer in S.O. and I think it could even get interesting/pedagogical ones here. I agree the question could be re-formulated to evince its statistical content, but I can also see how that could be hard for the OP (if my assumption is correct that the OP is missing part of the statistical content of the question). – user603 Feb 18 '14 at 19:40
  • @user603, it seems the question has already been closed, but I can't imagine it will be deleted. I am happy for the question to be here, if it is on-topic. You might see if you can initiate a conversation w/ the OP, if they come back, to help them re-formulate the question. – gung - Reinstate Monica Feb 18 '14 at 20:39
  • I also think this question contains more of a statistical matter than the usual "how to do X with R" as it seems the OP needs some Stats advice as well. @user40494 would you care to expand on your question a bit? What is the goal? Do you have a panel, a time series or cross-sectional data? What are your predictors? It's probably also a good idea to share the real or similar data to get a good answer. – Momo Feb 18 '14 at 21:06
  • My data set consists of automobile claims frequency and amounts. I have predictors like make,model,cc,age of vehicle,insured gender and age,geographical area..I want to fit a Tweedie Compound Poisson Gamma to my aggregate claims data in R. I want to know if its tractable and how to i proceed with it?@ Momo @ user603 – user40494 Feb 19 '14 at 05:14

0 Answers0