Objective

ABC Company has an OLTP database which keeps records on motor vehicle insurance information. This database contains detailed information in respect of drivers, vehicle and claim information.

Current database model has been designed for fast data entry and is sufficient for individual client’s specific information as well as fast transaction processing. Company critically needs to make comprehensive analysis like identification of contracts with a high loss ratio and low overall customer value.

Appropriate actions needed to be taken with high risk customers, such as premium adjustment, loss prevention measures and in some cases contract cancellations and reduce gross claim expenditures.

By making evidenced based management and informed decision, company will focus on profitable customers by lowering their premiums and overcoming competitive pressure. It will help company make better risk management and overall profitability for the company. ABC Company is in urgent need to utilize the existing data resources efficiently for better risk management and obtain competitive advantage in Auto Insurance Industry.

Recommended Solution;

ABC Company has decided to implement a Data Warehouse to leverage its data resources. ABC Company needs to reorganize the existing process of information delivery and to establish one single, unified and integrated data warehouse. A data warehouse is an integrated subject oriented, time-variant, non-volatile database that provides support for decision making.

In order to support decision making ABC Company decided to reorganize the data into Star Schema in Data warehouse. In effect, the star schema creates near equivalent of multidimensional database schema from the existing OLTP relational database [1]. It will help in advance data analysis for Risk management and overcoming competitive pressure.

Contd to Page-2
Page-2

Structure of Star Schema

Star schema yield an easily implemented model for multidimensional data analysis while still preserving the relational structure on which the operation database is built. [3] The basic star schema has four components: facts, dimensions, attributes and attributes hierarchies. The STAR schema would most likely be a read-only database due to the widespread redundancy introduced into the model. [4]

Fact Table

ABC Company has a factual data in Claim Information such as date, location, type of accident, cause of accident, liability, recovery cost.[5] Fact tables contain the quantitative data or factual data about a business. This information is numerical, additive measurements and can consist of many columns and millions or billions of rows.

Dimensions

Claim Information facts can be analyzed by dimensions such as Driver, Location, Time, and Automotive. Dimension tables are usually smaller and hold descriptive data that reflects the dimensions.

Attributes

For example Driver name, Driver ID, gender, age group, race, and other attributes. Some of these attributes might relate to each other hierarchically.

Attribute Hierarchies

Provide top down data Aggregation, Drill down or roll up data analysis. For example in time dimension there are Attribute hierarchies such as day, week, month, quarter, and year. When decision maker want to see company yearly claim information, then they are using year hierarchy level, they can further drill down to quarter level sales quantity, as per there needs. Same as in Location dimension is data can be analyzed by Country, Region, Province City and town.

Benefits of Data warehouse to ABC Company

By organizing the ABC Company data around star scheme company can analyze information like what customers are high risk and what group of customers is profitable. What cities have more accidents ratios and what time of the year accident happens? What habit of drivers is may be considered high risk? What vehicles are considered low risk and so on?
Contd – Page-3

Page-3

In addition to the internal information some external information like Auto Industry statistics can also be integrated into data warehouse.

By having answers to ad-hoc queries and in depth data analysis, ABC Company will be able to manage customer relations, smartly overcoming competitive pressure and most important of all is significantly improved risk management.

Data Warehouse will enable company have the business Intelligence for making strategic decision for Risk management and keep the company ahead of the competition and possibly diversify into new auto insurance products. [2]

References:

1- Data model overview

http://www.teradata.com/t/WorkArea/DownloadAsset.aspx?id=2332

2- The Benefits of Data Warehousing for Insurance Company Wolfgang Hofbauer, Mannheimer AG Holding, Mannheim, Germany

3- Database Systems By Peter Rob, Carlos Coronel, Keeley Crockett – Published by Thomson Learning; International Ed edition (12 Mar 2008)

4- Oracle Data Warehouse Tips by Burleson Consulting.

5- Class notes and research with Insurance companies.

Staff must agree on exactly what constitutes a “customer” or a “partner,” and how to resolve any disagreements across business units. Departments and divisions need to agree on hierarchies of customers and products and how to resolve duplicate records across sources.[1]

Source:
1- Demystifying Master Data Management by CIO.com

Bottom Up approach
Faster and easier implementation of manageable pieces of information. Inherently incremental and important data marts can be scheduled first as per end user requirements. Facilitate project team to learn and grow. Each data mart has narrow scope, data redundancy and inconsistency may be present.

The best approach
Best approach is the basic understanding of information needs of an organization. That must include both present and long-term information requirements. In view of that a careful implementation and monitoring is critical.

Dimension keys uniquely identify each record in a dimension table. The purpose of dimension keys is to relate one dimension table record to a fact table record. Thus, the dimension key must be stored in both the dimension table and the fact table.

We tried our best to elaborate some terms used in Business intelligence, Data Warehousing and Data Mart.
If readers have some more to add / please feel free to submit in comments so it can help other readers and correct ourselves.

Data Mart Structure

The data we used for business intelligence can be divided into four
categories: measures, dimensions, attributes, hierarchies.The four type of data help us to define the structure of data mart.[1]

Measures

Measure is numeric quantity expressing some aspect organization’s
performance. Measures are the facts and also known as fact table.[1]

Dimension

Dimension is categorization used to spread out aggregated measure to reveal its constituent parts.[1] dimensions include time (or date), customer, product, geography, lab type, campus, patient, promotions, gender (and other demographics), and so forth. Each dimension is associated with the facts / measures to which it relates via the linkages / joins between the table(s) housing the dimension (the dimension table) and the fact table.[2]

Attributes
For example a member of the Patient dimension, within the Analysis Services implementation for a healthcare provider, might contain information such as patient name, patient ID, gender, age group, race, and other attributes. Some of these attributes might relate to each other hierarchically, and, as we shall see in subsequent articles of this subseries (as well as within other of my articles), multiple hierarchies of this sort are common in real-world dimensions.[2]
Hierarchies

These are created from the attributes of dimesions. For example In Dimension Geography there are attributes named as Country-State-City-Postal Code.[3]

Source

1- Delivering Business Intelligence with Sql Server 2008 by brian Larson
2- Database Journal by Internet.com
3- Sql Server Analysis Services 2008 with MDX

This workshop in addition to giving a Data Mart Structure also give picture of Graduate admissions Star Schema. We find this a good example of Star Schema. Please visit this link to read more.

Health Care Claims Data Mart: Construction and Exploitation Marge Scerbo, CHPDM, University of Maryland Baltimore County Stuart B. Levine, SAS Institute Inc.

This paper will cover the development of a Data Mart to meeting the information and business intelligence needs. This includes the steps taken in the design of the Mart, its base tables and MDDBs (MultiDimensional DataBase), the definition of SAS/EIS objects that can exploit MDDBs, and other specific customizations that were developed to allow one organization, the UMBC Center for Health Program Development and Management (CHPDM), to address data access issues.

To read all please go to the website