Unlocking Your Data Treasury

1
Unlocking Your Data Treasury

• Capt Sean Sadlier
• LT Madalyn Spivey
• LT Warren Yu

2
Legacy Systems

• Accounting systems
• Management Information Systems
• Different systems for former agencies, different
  programs
• Mainframe extracts used for analysis
• Systems not integrated

3
Legacy Systems Problems

• Untimely reports and analyses
• Data hard to manipulate and analyze
• Databases linked by hand
• Skill required to extract and use data
• Data in hands of systems analysts, not
  policymakers and program managers

4
Whats Data-mining?

• Finds patterns and relationships in data by using
  sophisticated techniques to build models that
  represent reality.

5
Goals of Data Mining Project

• Make data available to all employees
• Transparent links across data sources
• Low cost of implementation
• Short learning curve for users
• Access and analysis capabilities scaleable across
  users

6
Information Needs

• Fund statuswho, what, where, when, how much?
• Portfolio servicingWhere are the problems? What
  trends and patterns?
• Targetingare we meeting our policy goals?
• ReportingCongress, OMB, GAO

7
Types of Data Users
System programmers (5)
Serious analysts (5-10)
Occasional, light data users (10-15)
Users of canned reports (75)
8
The Solution

• Leave legacy databases in place
• Populate corporate Data Warehouse
• Draw daily extracts into SQL database
• Integrate data elements at desktop level
• Make available to users via Web
• Give users ability to design custom reports

9
Benefits

• Uses current investment in data architecture
• Does not require additional client software
• Does not require staff training to use
• Enables user customizability
• Provides instant response

10
Problems Encountered

• Data problems
• Access problems
• Program limitations
• User requirements unclear
• Staff limitations
• System performance problems
• Political problems

11
Data Issues

• Illegal characters
• Incorrectly coded data
• Mismatched elements and definitions across data
  sources
• Large database tables

12
Access Issues

• Mainframe access and connection installed and
  configured for use
• DBA restrictions on data sources created copies
  of tables without DBA restrictions
• Staff access limitations
• Parameter passing
• Browser limitations

13
System Performance Issues

• Users need to define what capabilities they want
• Staff need to be skilled in data mining and
  internal computations
• Java-capable browsers
• PCs require speed, memory
• Bandwidth issues

14
Political Issues

• Job security
• Only one program area willing to provide
  program-savvy staff
• No initial buy-in from middle management
• Little interest in new ways of doing business
• Need to see working product to believe

15
Success Needs

• Confidence of Senior Management
• Support and buy-in from CIO
• Build immediate products to gain support
• Advocate value to employees
• Willingness to change directions based on early
  experience