Group Ins. - Data Sources And Data Structures

Front 1

Main focuses of this chapter:

Back 1

 Using Internal Data and External Data
 Data Quality Issues
 Data Structures (in the computer science sense)

There’s very little that is unique in this chapter. Most of it is covered in my General Notes.

Front 2

Responsibilities of the Actuary regarding usage of data:

Back 2

 Quote the entire General-Data Quality Issues.

Front 3

Uses Of External Data & Uses Of Internal Data

Back 3

 Quote the entire General-Data Sources: Internal vs. External Data

Front 4

Actuarial Information Systems (AIS’s)

Back 4

AIS’s are programs that allow the Insurance Company to analyze its internal data.
 AIS’s are designed to facilitate reporting rather than processing
 AIS’s allow hands-on manipulation of data
 They are often used for:
 computing reserves
 trend analysis (utilization, mortality, medical costs)
 experience monitoring
 ratesetting
 provider/network analysis (utilization, etc.)

Front 5

Data Storage Methods

Back 5

 On-line – means the file is located on disk or server storage that is immediately available for use.
 Near line – the file is compressed, but the compressed file is located on an immediately-available disk or server.
 this is cheaper than on-line storage
 it’s a compromise between online and offline: you have to decompress the data before you can use it, but that step is reasonably quick.
 overall, more efficient, because it takes less hard disk space or server disk space.
 Off-line – the file is located on a removable magnetic tape or floppy disk
 least expensive storage method
 but access is slowest.
 This storage method should only be used for sequential data which does not need to be accessed often. For example, a monthly backup file of the customer database.
 Trade-off: On-line is the most efficient, but requires the most expensive equipment and storage media. Off-line is cheapest but least efficient.

Front 6

Types of Data Structures

Back 6

 Sequential File – A simple list of all the data for a specific application
 e.g. a premium billing record, listing all the employer groups and their premiums.
To search for particular data in a sequential file, the computer must scan all the records until the desired record is found. Even if the data are listed in ascending order, searching is still slow.
 Indexed Sequential File – A list of data records with indexes of the key elements.
(The indexed elements are SSN, Policyholder Name, Contract #, etc; anything you would use to sort the data).
 Indexed data is easily searchable/sortable. Therefore it is often useful for more than one application.

 Relational Database – several indexed files linked together.
 data are grouped into separate indexed tables
 Given any piece of data, all related data from any of the tables can quickly be displayed
 Requires most storage space
Examples: Microsoft FoxPro, Microsoft Access, Oracle

 Dimensional Database
 Contains even more linkage than a relational database. (I think this is a type of fuzzy logic system)

 Trade-off: The more advanced kinds are efficient but most expensive. Sequential is cheapest but slowest.

 The Data Storage media and the Data Structure both affect the speed at which a desired piece of data can be accessed.

Front 7

Considerations in Choosing the Data Storage Method and Data Structure

Back 7

 The desired ability to manipulate (use) the data
 The available storage space
 Volume of Data
 Large data files need structuring in order to minimize processing time
 They also require more storage space; on-line storage isn’t possible.
 The desired frequency of use
 Frequently accessed files require relational databases, because sequential files take too much time to access.
 Retrieval Speed desired
 On-line is fastest to access
 Off-line data is slowest to access
 Near-line is a compromise
 Trade-Off: Ease of Programming vs. Accessibility.
 Complex data structures are harder to create, but allow data to be accessed in a number of ways. They are more efficient in the long-run.

Front 8

Sources of Internal Data

Back 8

 Quote the entire General-Data Sources: An MCO’s Internal Data Sources.


Done.