
#151




Quote:
...but... Isn't today the deadline to register? Doesn't seem like you have much time to decide!
__________________
ASA Countdown Exams: FAP: 
#153




Quote:

#156




Quote:
__________________
Spoiler:  Want to connect on LinkedIn? PM me! 
#157




Dec exam syllabus: “The December 2019 exam will be structured in a manner similar to this sample project and to the June 2019 exam.“

#158




I'm trying to go through and build my own GLMs for the sample project in RMD6.8. I don't understand the one where they build the FACEratio gamma/log link glm. I initially didn't include logINCOME as I assumed since the target is FACE/INCOME that I couldn't use this variable. When running the model without logINCOME, I get an error, inner loop 1 cannot correct step size. When I add the variable, I get a warning glm.fit: algorithm did not converge.
I would appreciate any help explaining this since the solution RMD basically glosses over it. Edit: I also noticed I didn't include SAGE since AGEdiff is a linear combination of AGE and SAGE. Why would they be including this as well?
__________________
VEE: FAP: FSA Mods: Favorite Beer: M43 Last edited by Chappy; 11152019 at 02:12 PM.. 
#159




I don’t know how to dig into the convergence issue for multi variables, and last sitting’s solution made it seem like being able to explain the issue conceptually is good enough. It should converge if you use Gamma’s canonical link function, which is the negative inverse. Tradeoff is poor interpretability of the predictors’ coefficients.

#160




Quote:
I just purchased the online version of the material. Where do I go to get started on this. I am little be late in the game so please I am needing this answer as quickly as possible.
__________________
Prelim: VEE: FAP 
Thread Tools  Search this Thread 
Display Modes  

