

FlashChat  Actuarial Discussion  Preliminary Exams  CAS/SOA Exams  Cyberchat  Around the World  Suggestions 

ShortTerm Actuarial Math Old Exam C Forum 

Thread Tools  Search this Thread  Display Modes 
#551




I've read a lot of varying opinions on this exam. Some people say doing problems wont work and you have to know the material like the back of your hand. Some other people say they just did a ton of practice exams and ended up passing with flying colors. I'm hoping enough practice problems from different sources will help me enough. I'm getting nervous...
__________________
ASA FSA Exams: FSA Modules: Reg & Tax  
#557




lol just watched TIA's 'before you do the sample exams' strategy guide, and Dave said something like "...you don't go out drinking the night before the exam, that's something you do after the exam"

#558




Quote:
I suspect what you mean is that you noticed that the prior distribution is a 40/60 mixture of two Gamma random variables. It does indeed turn out that the posterior distribution will be SOME mixture of Gamma random variables, namely of the two Gamma distributions that would be the posterior distributions that would result solely from each of the Gamma distributions in the mixed prior. But this final mixture isn't the same as the original 40/60in order for it to be so, the posterior distribution would have to result from a linear computation in the density function, and that fails to be true because of the division in the computation.
__________________
Jim Daniel Jim Daniel's Actuarial Seminars www.actuarialseminars.com jimdaniel@actuarialseminars.com 
#559




Holy crudsticks am I getting destroyed by practice exams, lol.
EDIT: This test is going to own my soul so hard.
__________________
Last edited by zenkei18; 06062016 at 04:22 PM.. 
#560




So, when we are dealing with bayesian credibility with normal/normal prior conjugate, and we have lognormal as the model distribution, we keep the model variance the same and just LN the sample losses, correct? So basically, when we have a lognormal distribution instead of normal in this type of problems, only portion that i have to change is the sample mean.

Thread Tools  Search this Thread 
Display Modes  

