Actuarial Outpost
 
Go Back   Actuarial Outpost > Search Forums
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions


Showing results 1 to 25 of 206
Search took 0.02 seconds.
Search: Posts Made By: TranceBrah
Forum: Life Pricing Exam 07-25-2019, 01:06 AM
Replies: 6
Views: 1,283
Posted By TranceBrah
I just read this now and it seems like they are...

I just read this now and it seems like they are both using the conservation of death to calculate the persister mortality. The difference is that B-K is just the same as D-M but with the extra factor...
Forum: Exam PA: Predictive Analytics 07-23-2019, 02:45 PM
Replies: 12
Views: 1,210
Posted By TranceBrah
15 days, 1.8k raise, 3.6k bonus (Canada). ...

15 days, 1.8k raise, 3.6k bonus (Canada).
Slightly reduced days for 2nd and 3rd attempt
Forum: Life Pricing Exam 07-16-2019, 11:55 PM
Replies: 6
Views: 1,283
Posted By TranceBrah
I'm not scheduled to be there for another 9 days!

I'm not scheduled to be there for another 9 days!
Forum: Life Pricing Exam 06-21-2019, 01:10 PM
Replies: 12
Views: 1,614
Posted By TranceBrah
What does it mean when the material is released...

What does it mean when the material is released over time? Are people who purchased the electronic version getting updated pdfs? How about those that went for the printed version?
Forum: Life Pricing Exam 06-20-2019, 08:57 PM
Replies: 6
Views: 1,283
Posted By TranceBrah
Fall 2019 Progress Thread

Let's get that bread together! Going to buy all the study material this weekend.
Forum: Exam PA: Predictive Analytics 06-18-2019, 12:45 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
1 and 2 were both 5 points

1 and 2 were both 5 points
Forum: Exam PA: Predictive Analytics 06-16-2019, 10:46 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
lol i cant even tell if the two days had the same...

lol i cant even tell if the two days had the same questions anymore! clearly different data since the RMSE were different.

btw, did you guys mention the aggregated methods in that tree question?...
Forum: Exam PA: Predictive Analytics 06-15-2019, 02:48 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
ok I saw the same thing for 1st component, but u...

ok I saw the same thing for 1st component, but u can see the magnitude of coefficients werevery large for other variables in the other components, so how did you decide on a coeffcient?



went...
Forum: Exam PA: Predictive Analytics 06-15-2019, 11:08 AM
Replies: 645
Views: 100,152
Posted By TranceBrah
A few things that shook me up were.. - #2...

A few things that shook me up were..

- #2 combining the factor levels took a long ass time because u had to do it for like 8 variables.... and comment on the reasons for all of them

- #3 the...
Forum: Exam PA: Predictive Analytics 06-13-2019, 06:11 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
have a couple of concerns but from the ppl i...

have a couple of concerns but from the ppl i talked to outside the room they had the same results as me... will wait til we are allowed to discuss.
Forum: Exam PA: Predictive Analytics 06-13-2019, 05:22 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
im shook from that

im shook from that
Forum: Exam PA: Predictive Analytics 06-12-2019, 05:57 PM
Replies: 45
Views: 4,078
Posted By TranceBrah
Agree Totally off the mark with this one. It...

Agree

Totally off the mark with this one. It has nothing to do with your distribution.

Offset is a predictor variable with known coefficient of 1. Think of this as modifying your base level...
Forum: Exam PA: Predictive Analytics 06-12-2019, 02:03 AM
Replies: 80
Views: 7,160
Posted By TranceBrah
This is exactly what u want to do. In fact,...

This is exactly what u want to do. In fact, comparing on the training set is what u should consider cheating :)
Forum: Exam PA: Predictive Analytics 06-12-2019, 01:27 AM
Replies: 645
Views: 100,152
Posted By TranceBrah
The p-value for calculated for each level tells...

The p-value for calculated for each level tells you if that level is significantly different from your base level (recall your glm does not assign a coefficient to the base level).

Imagine a...
Forum: Exam PA: Predictive Analytics 06-11-2019, 11:00 AM
Replies: 45
Views: 4,078
Posted By TranceBrah
It's funny because I was thinking about this a...

It's funny because I was thinking about this a month ago. And basically all the Googling will lead to the conclusion that using stepAIC, drop1, or removing p-values are all terrible ideas (and even...
Forum: Exam PA: Predictive Analytics 06-11-2019, 10:50 AM
Replies: 13
Views: 810
Posted By TranceBrah
I would research and give a good answer to this...

I would research and give a good answer to this thread but at this point I realize it doesn't matter.

Try 0.10, too simple. Try 0.001, too complex. PlotCP says use 0.10. Looks pretty good. Next...
Forum: Exam PA: Predictive Analytics 06-11-2019, 02:39 AM
Replies: 645
Views: 100,152
Posted By TranceBrah
major key right here :toast:

major key right here :toast:
Forum: Exam PA: Predictive Analytics 06-09-2019, 03:37 AM
Replies: 645
Views: 100,152
Posted By TranceBrah
yep they would be equivalent. A) y = a + b +...

yep they would be equivalent.

A) y = a + b + a*b + c
B) y = a + b + a:b + c
C) y = a*b + c

Will all result in the same coefficients.
Forum: Exam PA: Predictive Analytics 06-09-2019, 02:40 AM
Replies: 645
Views: 100,152
Posted By TranceBrah
the difference is that y~a*b = y~a+b+a:b, so your...

the difference is that y~a*b = y~a+b+a:b, so your glm will produce 3 coefficients, not counting intercept.
y~a:b would just give you that one coefficient. In the Dec exam they did it specifically...
Forum: Exam PA: Predictive Analytics 06-08-2019, 08:54 PM
Replies: 45
Views: 4,078
Posted By TranceBrah
actually i'm not that confident with pdp and...

actually i'm not that confident with pdp and hierarchical (although the code seems like 1 liners that would be provided anyways).

gonna just focus on writing style these next few days
Forum: Exam PA: Predictive Analytics 06-08-2019, 07:31 PM
Replies: 45
Views: 4,078
Posted By TranceBrah
anybody able to find a nice example of tweedie...

anybody able to find a nice example of tweedie being implemented (within glm function)? mostly just seeing results from people with problems (eg. coefficients same as gamma)
Forum: Exam PA: Predictive Analytics 06-08-2019, 06:17 PM
Replies: 45
Views: 4,078
Posted By TranceBrah
huh. that's nice to know. 1 less thing to worry...

huh. that's nice to know. 1 less thing to worry about.
Forum: Exam PA: Predictive Analytics 06-08-2019, 01:41 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
Looks like you know exactly how to score a 10...

Looks like you know exactly how to score a 10 :toast:
Forum: Exam PA: Predictive Analytics 06-08-2019, 01:30 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
Not sure about "inputs" but yeah generally PCA is...

Not sure about "inputs" but yeah generally PCA is just a tool for dimension reduction.


This could be a valid approach but most likely a waste of time IMO. I think the main reason in practice is...
Forum: Exam PA: Predictive Analytics 06-08-2019, 12:19 PM
Replies: 645
Views: 100,152
Posted By TranceBrah
If you mean target when you say "predicted value"...

If you mean target when you say "predicted value" then yes.

Typically you dont use PCA on all your predictors, only a subset. So you would do something like prcomp(data[,c(5,9,13)]) to only use...
Showing results 1 to 25 of 206

 

All times are GMT -4. The time now is 10:43 PM.


Powered by vBulletin®
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
*PLEASE NOTE: Posts are not checked for accuracy, and do not
represent the views of the Actuarial Outpost or its sponsors.
Page generated in 0.25246 seconds with 7 queries