Presentation is loading. Please wait.

Presentation is loading. Please wait.

Building a Data Warehouse for Business Reporting Presented by – Arpit Desai Faculty Advisor – Dr. Meiliu Lu CSC Department – Spring 2006 California State.

Similar presentations


Presentation on theme: "Building a Data Warehouse for Business Reporting Presented by – Arpit Desai Faculty Advisor – Dr. Meiliu Lu CSC Department – Spring 2006 California State."— Presentation transcript:

1 Building a Data Warehouse for Business Reporting Presented by – Arpit Desai Faculty Advisor – Dr. Meiliu Lu CSC Department – Spring 2006 California State University, Sacramento

2 Agenda Fundamentals of Data Warehousing Fundamentals of Data Warehousing Data warehousing and Business Analysis Data warehousing and Business Analysis Total Loss Valuation Reporting (TLV)– An example project Total Loss Valuation Reporting (TLV)– An example project Designing the solution data mart Designing the solution data mart Implementing the design Implementing the design Business Reports for TLV Business Reports for TLV Performance enhancement techniques Performance enhancement techniques Questions & Suggestions Questions & Suggestions

3 Fundamentals of Data Warehousing (DW) What is OLTP?What is OLTP? OLTP – A technology that uses highly normalized tables to quickly record large number of transactions while making sure that these updates of data occur in as few places as possible OLTP – A technology that uses highly normalized tables to quickly record large number of transactions while making sure that these updates of data occur in as few places as possible What is OLAP?What is OLAP? OLAP – A technology that uses database tables to enable multi-dimensional viewing, analysis and query large amounts of data OLAP – A technology that uses database tables to enable multi-dimensional viewing, analysis and query large amounts of data What is a Data warehouse?What is a Data warehouse? DW is a collection of historic data from different functional operations of the company DW is a collection of historic data from different functional operations of the company What is a Data Mart?What is a Data Mart? Data Mart (DM) is a segment of DW that provides data for reporting Data Mart (DM) is a segment of DW that provides data for reporting

4 Data Warehousing and Business Analysis Benefits of DW Benefits of DW Useful information derived from operational dataUseful information derived from operational data Specially designed to quickly execute aggregate queries on large amounts of dataSpecially designed to quickly execute aggregate queries on large amounts of data Queries, complex in highly normalized databases, could be easier to build and maintain in data warehouses, decreasing the workload on transaction systemsQueries, complex in highly normalized databases, could be easier to build and maintain in data warehouses, decreasing the workload on transaction systems Business Requirements Business Requirements Organizations need to keep track of market trends – generally through numbersOrganizations need to keep track of market trends – generally through numbers Business analysts rely on Reports (facts, figures) to make critical business decisionsBusiness analysts rely on Reports (facts, figures) to make critical business decisions New features need to be added to existing products – like reportingNew features need to be added to existing products – like reporting

5 Total Loss Valuation (TLV) Reporting Sub-part of Mitchell WorkCenter tm, a product of Mitchell International, San Diego, CA Sub-part of Mitchell WorkCenter tm, a product of Mitchell International, San Diego, CA Vehicle damaged beyond repair is termed as “Totaled Out” Vehicle damaged beyond repair is termed as “Totaled Out” Insurance Company needs to analyze amount of money paid out Insurance Company needs to analyze amount of money paid out Summarized and Detailed reports about claims filed are required Summarized and Detailed reports about claims filed are required Search criteria consists of company hierarchy, types of coverage, date ranges, etc. Search criteria consists of company hierarchy, types of coverage, date ranges, etc.

6 Players of Auto Insurance Industry Claimant brings damaged car to the body shop Body Shop Adjuster creates a Valuation Request (estimate) If cost of repair is more than market price of the vehicle Then The vehicle gets Totaled Out Insurance company pays Settlement Value

7 Designing the Solution Data Mart (TLV) Functionally restricted to TLV Reports Functionally restricted to TLV Reports Operational data Operational data Data Sources/Formats standardizationData Sources/Formats standardization Data LoadData Load Initial load Initial load Periodic Updates Periodic Updates Identifying Dimensions Identifying Dimensions Define metrics values and aggregate amounts to be included in the Fact table Define metrics values and aggregate amounts to be included in the Fact table Design supporting structures like indexes, synonyms, database links, materialized views, global temp tables, foreign references Design supporting structures like indexes, synonyms, database links, materialized views, global temp tables, foreign references Design the complete Data Model with schema of execution Design the complete Data Model with schema of execution

8 Implementing the Design (TLV) Extract, Transform & Load Extract, Transform & Load Initial Load involves Initial Load involves Staff DimensionStaff Dimension All user who ever created a Valuation request – Adjuster All user who ever created a Valuation request – Adjuster Office DimensionOffice Dimension All Offices and company hierarchy, along with its geo-location details All Offices and company hierarchy, along with its geo-location details Valuation Info DimensionValuation Info Dimension All the Valuation requests created thus far and its details All the Valuation requests created thus far and its details Time DimensionTime Dimension All possible dates and summarizations All possible dates and summarizations Valuation History DimensionValuation History Dimension Currently not loaded, reserved for future extension Currently not loaded, reserved for future extension Valuation FactValuation Fact Contains aggregate values, counts, reference keys Contains aggregate values, counts, reference keys Periodic Update Periodic Update Data Mart refreshes nightly – flushes all the tables and reloads (poor design)Data Mart refreshes nightly – flushes all the tables and reloads (poor design)

9 Business Reporting (TLV) Claim Detail Report Procedure Claim Detail Report Procedure Receives selection criteria as input parameters and returns a reference cursorReceives selection criteria as input parameters and returns a reference cursor Shows details of all the claims submitted, which meet the selection criteriaShows details of all the claims submitted, which meet the selection criteria Claim Summary Report Procedure Claim Summary Report Procedure Input report selection criteria output is reference cursorInput report selection criteria output is reference cursor Shows summarized counts, averages, percentages for claims which meet the selection criteriaShows summarized counts, averages, percentages for claims which meet the selection criteria

10 Performance enhancement techniques Use Star schema for query execution and Bitmap indexes for all tables Use Star schema for query execution and Bitmap indexes for all tables Nightly refresh of data mart Nightly refresh of data mart Use ‘truncate’ instead of ‘delete’Use ‘truncate’ instead of ‘delete’ ‘Rebuild’ indexes instead of ‘drop-recreate’‘Rebuild’ indexes instead of ‘drop-recreate’ Run report procedures against views, offers flexibility to data mart structure Run report procedures against views, offers flexibility to data mart structure Build and unit test each chunk of extraction queries separately Build and unit test each chunk of extraction queries separately

11 Sample Queries that the data mart processes Select a.item1, a.item2, …, b.item1, c.item1, c.item2 from xxx_fact a, xxx_dim b, xxx_dim c wherea.id_1=b.id_1 and b.id_2=c.id_2 and b.item5 = and b.item5 = and a.item6 = and a.item6 = and c.item9 = and c.item9 = Order by c.item1

12 Questions & Suggestions Thank you


Download ppt "Building a Data Warehouse for Business Reporting Presented by – Arpit Desai Faculty Advisor – Dr. Meiliu Lu CSC Department – Spring 2006 California State."

Similar presentations


Ads by Google