Transcript Business Intelligence on a Budget Presentation (Slides)

Business Intelligence
on a Budget
•
•
•
•
•
Born in Texas, Raised in Oklahoma
Undergrad from OU
MBA from Texas A&M
Married with 4 Kids, 2 Boys and 2 Girls
Ranging from 19 Years Old to 7 Years Old
• 7 Years Working at Collin College
• Before That, 2 Years at the University of Central Oklahoma
• 14 Years of Combined Experience with BI, CRM, and SFA
Solutions
About Me
Christopher Frost
About Me
My Life as a Series of Graphs
Kids
25
20
15
10
5
0
Years Till
College
Graduation
Years Old
Jobs
Collin
UCO
UPS
MCI
AA
About Me
My Life as an Infographic
• What stories does your data tell about you?
• Digital Footprint
• What stories do you want data to tell you?
• Student success factors
• Outcome metrics
• Interest in programs and courses offered
About Data
Drowning in a Sea of Data
PROS
• Central Repository of
Information
• Secure
• Backed Up
• Available to
Authorized Users
CONS
• Data Might Be Hard
to Access
• Limited QA
• IT Might Be Tracking
Data You Don’t Need
& Ignoring Data You
Do Need
About Data
IT as a Librarian
• Wikileaks
• School Performance
http://schools.chicagotribune.com/
• Comparative School Data
http://projects.propublica.org/schools/
About Data
Power of Stories
• End users are more sophisticated
• Want more access to the raw data
• Want to tell their own stories
About Data
Rules are Changing…
…to this
About Data
Moving from this…
• “I don’t know what I want, but I’ll know it when I see it.”
• People are wired to react to visuals
• Things are moving from static reports to interactive
dashboards
• OLAP cubes allow users to dig into the total numbers and
look at underlying trends
• This ability to dig deeper allows us to tell better, more
accurate stories.
About Data
From Reports to Dashboards
• Pentaho
• Microsoft’s SharePoint / SSRS
• Oracle’s OBIEE / Oracle Answers
• SPSS
• Cognos
Data Tools
Sophisticated BI Systems
• Tableau Public
Data Tools
Budget-Minded BI Solutions
• Did anyone bring a spreadsheet to share?
SHORTCOMINGS OF TABLEAU PUBLIC
• Free version means the data is public for
everyone.
• Don’t share if you care about privacy.
• Limited to 100,000 rows.
Data Tools
Tableau Public Demo
• Tableau Public
• ???
Data Tools
Budget-Minded BI Solutions
• Is the One Used Most Often - Microsoft Excel
… and PowerPivot Can Make It Much Better
Data Tools
Most Powerful BI Tool
• Tableau Public
• Microsoft Excel
• PowerPivot
Data Tools
Budget-Minded BI Solutions
Data Tools
Setting Up PowerPivot
Data Tools
Setting Up PowerPivot
Data Tools
Setting Up PowerPivot
• Work with Large Data Sets
• Combine Data from Different Data Sources
• However Only Inner Joins Allowed
• Easy to Set Up Dashboard-Style Views
• Slicers Make It Easy to Visually Filter Data
• Use DAX to Create Calculated Columns on the Fly
• Working with Dates (i.e. YEAR(H2))
Data Tools
PowerPivot Features
Data Tools
PowerPivot Features
Data Tools
PowerPivot Features
• Did anyone bring a spreadsheet to use?
Data Tools
PowerPivot Demo
• With the Data Tab and Get External Data
Data Tools
How Can I Get to the Data?
• Can Use Excel to Access the Following Data:
• Access Databases
• SQL Server Databases
• CSV Files
• Text Files
• XML Files
Data Tools
How Can I Get to the Data?
• Can Also Use Excel to Query Data in:
• Oracle Databases
• mySQL Databases
However, They Require Special
Drivers and Client Installs Before
They Can Be Used
Data Tools
How Can I Get to the Data?
• Metadata is the data about your data
• Can use this as a roadmap to identify those
elements that you are interested in
• Metadata already exists inside of your database
as column comments, datatypes, and table
comments.
• Best to document metadata as you create new
tables or modify existing ones.
Data Tools
Mining the Metadata
• Tableau Public
• Microsoft Excel
• PowerPivot
• Oracle Data Modeler 4.0
Data Tools
Budget-Minded BI Solutions
• Oracle Data Modeler 4.0 Demo
Data Tools
Mining the Metadata
End Result:
• Glossary of Available Data Sources
• Security Defined via an Object’s Associated Roles
• Data Element Dictionary
BEST OF ALL
• It’s easily updatable since it sits along side your
data.
Data Tools
Mining the Metadata
• Biggest issue isn’t technical. It’s functional.
• Data owners are justifiably protective of their
area’s data.
• Need assurances and safeguards in place to
prevent data from being misused.
Data Processes
Roadblocks to Data Access
For Example:
• Student needs assurances that Personally
Identifiable Info isn’t going leak out
• Finance needs confidentiality on cost center
authorizations
• HR needs assurances that performance-related
comments don’t become public
Data Processes
Roadblocks to Data Access
• IT works with a wide variety of areas
• Can facilitate two-way conversations between
departments on shared data
• What data do you want?
• What data are you willing to share?
• What concerns do you have about sharing data
with others?
• Start small and build trust between user
communities over time.
Data Processes
IT as Matchmaker
• Can Set Up Views of the Data that Filter Out
Sensitive Info
• Use a Surrogate Key for Granularity
• Associate Views with Security Appropriate Roles
and User Accounts
Data Processes
Balancing Access with Security
PUBLIC
ADMIN ASSISTANTS
DEANS
VPs
Registration Reports
Enrollment Reports
Course Schedule
Section Info
Outcomes and Retention Info
Program Effectiveness
Early Warning Systems
Learner Info
Access to Everything
Data Processes
Balancing Access with Security
• Can Set Up Views of the Data that Filter Out
Sensitive Info
• Use a Surrogate Key for Granularity
• Associate Views with Security Appropriate Roles
and User Accounts
Data Processes
Balancing Access with Security
• Views against Source Data and Working with
Spreadsheets will only Get You So Far
• Building a Data Warehouse Takes Planning, But
Isn’t Rocket Science
• Data Warehouse = Source Data through Time
• Clumping Data Together in Denormalized
Structures
• Basically, Turning 4000+ Tables into a Dozen Tables
that Fit Together
Data Warehousing
Build a Data Warehouse?
• Greater Performance since Totals are Often Preaggregated at a Granular Level
• Tables Can Be Optimized via Range or Hash
Partitioning
• Calls to the Source System are Reduced by
Having a Reporting System Available
Data Warehousing
Advantages
Data Warehousing
Data Warehousing
Data Warehousing
Data Warehousing
Data Warehousing
Data Warehousing
• Facts are your numeric data, your measures
• Level of detail on the fact table is the grain
• Facts are joined to dimensions in the star
schema
• Dimensions provide context for facts
• They are used for filtering queries or reports
• Dimensions control the grouping of facts when
you run the totals
Data Warehousing
Data Warehousing
• How will you deal with slowly changing
dimensions?
• Will you..
• Overwrite the data
• Or load the new value while “inactivating” the old
value
• Most have a mixture, depending on the
importance of that historical info.
Data Warehousing
Questions to Ask
• Start Small, Add On When You Can
• Avoid Inner Join Conditions
• Be Careful with NULL Values
• Look at Count Comparisons and Time Values to
Validate the Accuracy and Timeliness of the Data
• Don’t Be Afraid to Start Over. Things Change and
Sometimes the Data Warehouse Needs to
Reflect That
Data Warehousing
Lessons Learned
• The two great masters of data warehousing are
Ralph Kimball and William Inmon. Any of the
books below are great starting points.
Final Points
Further Reading
• I would also highly recommend Lawrence Corr
and Chris Adamson. Both represent the best of
the new ideas in data warehousing.
Final Points
Further Reading
• Tim Smith teaches a great BI Certificate Course
at our Courtyard campus. However, there are
lots of good BI programs out there.
Final Points
Or Training
• Have a great day!
• A copy of this presentation will be available on
www.savvybi.com.
Final Points
Thanks for Attending