I want to apply the knowledge of this paper (Bayesian estimation of probabilities of default for low default portfolios, by Dirk Tasche) in R, but I can't find the right bayesian package and functions to use. Can you help me?

  • $\begingroup$ What functionality are you looking for? $\endgroup$
    – Bob Jansen
    Jun 21 '20 at 10:11
  • $\begingroup$ I am not sure if I understood your question correcly. I use R version 3.5.1. I cannot do computations for the depedent case ( like Table 2- Page 15) $\endgroup$
    – Jimmy
    Jun 21 '20 at 13:34
  • $\begingroup$ Hi @Jimmy, you should describe what was done in the paper as formulas. Some forms have analytic solutions, others do not. There are packages for those with analytic forms and for those without. You should also discuss your prior distribution. It may help to explain how you chose it as well. Because Bayesian methods are not "one-size-fits-all" the way a t-test is, you have to adapt the packages and tools yourself in R. $\endgroup$ Jun 22 '20 at 1:25
  • $\begingroup$ @Jimmy also, be careful what type of distribution you are working with. For example, are you working with a posterior distribution or a posterior predictive distribution? Also, are you looking for a point estimate or the entire posterior or entire predictive distribution? If you are looking for a point estimate, you should explain your loss function. It will change your software. $\endgroup$ Jun 22 '20 at 1:30
  • $\begingroup$ @Jimmy also, if you have never used a Bayesian method before and, especially if you lack formal training, then you should get a copy of Bolstad's "Introduction to Bayesian Statistics." smile.amazon.com/… $\endgroup$ Jun 22 '20 at 1:33

If I understand correctly, you're looking for an implementation of Monte Carlo simulations in this paper? If you can't find anything on CRAN or elsewhere on the internet, I think your best bets, in order are:

  1. Contact the author
  2. Implement it yourself
  • $\begingroup$ Thank you for your advice $\endgroup$
    – Jimmy
    Jun 21 '20 at 14:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.