Actuarial Outpost
 
Go Back   Actuarial Outpost > Actuarial Discussion Forum > Property - Casualty / General Insurance
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions


Fill in a brief DW Simpson Registration Form
to be contacted when new jobs meet your criteria.


Reply
 
Thread Tools Search this Thread Display Modes
  #1  
Old 10-15-2019, 07:57 PM
cokiwah cokiwah is offline
Member
CAS
 
Join Date: Aug 2012
Posts: 312
Default How to implement binned/combined variables into rating plan?

1) I am using an indicator variable in a GLM (i.e. 1 or 0) to capture a variable that is 0,1,2,3. I'm doing this because of thin data at 2 and 3.

To implement into rating plan, is it as simple as doing exp(coefficient*2) for 2 and exp(coefficient*3) for 3?


2) Another question, I am using a sum of frequencies as a variable. Each individual term in the sum relates to a cause of loss. Do I use the coefficient that comes out of the GLM for each of the cause of losses?
__________________
ACAS 7 8 9
Reply With Quote
  #2  
Old 10-16-2019, 10:10 AM
ALivelySedative's Avatar
ALivelySedative ALivelySedative is offline
Member
CAS
 
Join Date: Dec 2013
Location: Land of the Pine
College: UNC-Chapel Hill Alum
Favorite beer: Red Oak
Posts: 3,429
Default

IANA-GLM expert, but:

1) sounds wrong...seems like you should just use '0' and '1+', or judgmentally pick something for 2 and 3 that makes sense despite the thin data if you must have explicit values.

2) my understanding here is that you should be doing separate GLMs for the different perils (how do you define an exposure base if multiple perils are modeled at once?), then combining those GLM results/selections after the fact. I think CAS monograph 5 addresses this to some degree with respect to homeowners in particular. I've also heard detailed conversation at a couple conferences with respect to GLM by-peril rating that implies this as well.

Full disclosure: anyone else's answer here is probably better than mine.
__________________
Stuff | 6 | ACAS | FCAS stuff
Reply With Quote
  #3  
Old 10-16-2019, 10:13 AM
#Hawaii2019's Avatar
#Hawaii2019 #Hawaii2019 is online now
Member
CAS
 
Join Date: Sep 2017
Studying for Hawaii
Posts: 837
Default

Quote:
Originally Posted by ALivelySedative View Post

1) sounds wrong...seems like you should just use '0' and '1+', or judgmentally pick something for 2 and 3 that makes sense despite the thin data if you must have explicit values.
Reply With Quote
  #4  
Old 10-16-2019, 10:17 AM
#Hawaii2019's Avatar
#Hawaii2019 #Hawaii2019 is online now
Member
CAS
 
Join Date: Sep 2017
Studying for Hawaii
Posts: 837
Default

Quote:
Originally Posted by ALivelySedative View Post

2) my understanding here is that you should be doing separate GLMs for the different perils (how do you define an exposure base if multiple perils are modeled at once?), then combining those GLM results/selections after the fact. I think CAS monograph 5 addresses this to some degree with respect to homeowners in particular. I've also heard detailed conversation at a couple conferences with respect to GLM by-peril rating that implies this as well.
also
Reply With Quote
  #5  
Old 10-16-2019, 10:43 AM
Vorian Atreides's Avatar
Vorian Atreides Vorian Atreides is offline
Wiki/Note Contributor
CAS
 
Join Date: Apr 2005
Location: As far as 3 cups of sugar will take you
Studying for ACAS
College: Hard Knocks
Favorite beer: Most German dark lagers
Posts: 66,534
Default

Quote:
Originally Posted by cokiwah View Post
1) I am using an indicator variable in a GLM (i.e. 1 or 0) to capture a variable that is 0,1,2,3. I'm doing this because of thin data at 2 and 3.

To implement into rating plan, is it as simple as doing exp(coefficient*2) for 2 and exp(coefficient*3) for 3?


2) Another question, I am using a sum of frequencies as a variable. Each individual term in the sum relates to a cause of loss. Do I use the coefficient that comes out of the GLM for each of the cause of losses?
How thin is that thin data? Are talking those levels having 15 claims across all perils? Do you get reasonable results but have extremely wide confidence intervals if you model them as separate levels?

What do you get if you still fit a linear (i.e., polynomial) model to this variable?
__________________
I find your lack of faith disturbing

Why should I worry about dying? Itís not going to happen in my lifetime!


Freedom of speech is not a license to discourtesy

#BLACKMATTERLIVES
Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


All times are GMT -4. The time now is 09:38 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 0.23119 seconds with 9 queries