Insurance Datawarehouse Model

raj.dwbi · Post by **raj.dwbi** » Thu Oct 02, 2008 8:17 am

hi everyone,
I'm currently assigned to a car insurance project. my previous domain was banking. can anyone help me in knowing about the basic architecture of the datawarehouse? Thanks

ArndW · Post by **ArndW** » Thu Oct 02, 2008 8:24 am

In most client scenarios the actual Model used to build the warehouse is a home-grown solution and often reflects the company's way of looking at and using data. There are a couple of different basic business-specific models that are available for purchase, but even these are usually highly modified by the time they come into production.

This means that your question cannot be answered here in the forum, but you should get a model description from your current site and work your way through that.

chulett · Post by **chulett** » Thu Oct 02, 2008 8:44 am

I'm sorry, but how can anyone here help you? Get 'the basic architecture of the datawarehouse' from the client, as Arnd notes I seriously doubt it's an out-of-the-box solution.

mikegohl · Post by **mikegohl** » Thu Oct 02, 2008 9:14 am

IBM has kind of a generic architecture. Here is the link:

http://t1d.www-03.cacheibm.com/industri ... un2006.pdf

Maybe this will help.

WoMaWil · Post by **WoMaWil** » Thu Oct 02, 2008 12:24 pm

We use it in our insurance company since some years and we fill it with DataStage. It is usefull but still in MS-Access instead of the Infosphere Repository.

kduke · Post by **kduke** » Thu Oct 02, 2008 4:59 pm

You have at least a couple fact tables to worry about. I would say premiums and claims would be the first fact tables. If you can get with underwriting and figure out how they rate then your fact tables need to track the data elements to validate the rating formulas. I would say some dimension would represent these elements like age and sex categories. They may rate by marital status. Some rate by credit ratings. All of these are how you want to dice and slice your information or facts. So they need to be dimension tables. Most insurance companies are concerned about marketing almost as much as financial infomation. This may add dimensions which may not be concerned about ratings formulas. Maybe they are more concerned about sales channels. Maybe a lot of your business is through the web. That is a whole other problem about what data elements are important. Maybe each of these is a departmental datamart. Maybe if you planned for conformed dimensions which can be used in all these departmental solutions then you have done your job well.