1) The document outlines an agile methodology for data analytics that focuses on collaborating with users to understand their goals and iteratively developing simple solutions to address critical needs, rather than extensive upfront documentation, planning, and separate teams.
2) In a case study, a traditional methodology required multiple rounds of rework when user needs changed, while an agile approach allowed rapidly loading data, running queries, and refining the solution with user feedback to group data differently.
3) Key aspects of the agile data analytics methodology include focusing on the user's value, collaborative requirements, simple and adaptable architecture, integrated cross-functional teams, and an attitude of accepting change.
FAIR, FAIRsharing, FAIR Cookbook and ELIXIR - Sansone SA - Boston 2024
Agile Data
1. Agile Manifesto - 2001
1
We are uncovering better ways of developing software by doing it
and helping others do it.
Through this work we have come to value:
Individuals and interactions over processes and tools
Working software over comprehensive documentation
Customer collaboration over contract negotiation
Responding to change over following a plan
That is, while there is value in the items on the right, we value the
items on the left more.
2. Agile Data Analytics
2
Case Study:
Demonstrate the value of a drug
prescription cost reduction program to
determine program pricing structure
and set the stage for an RFP analysis
tool to be used for contracting.
3. Traditional Methodology
3
End user reviews information
from report and realizes that
more data is needed.
Can we look at this by drug class
or maybe by provider specialty?
1. Gather business requirements: cost, price, by group
2. Identify source data: claims file
3. Design dimensional model: claims, group, calendar
4. Map source to DW tables
5. Build and test ETL
6. Design reports
7. Map DW to report fields
8. Build and test reports
Repeat cycle
4. Round Two
4
End user reviews information
from report and realizes that
more data is needed.
Specialty on the claims is pretty
sketchy, can you use this
directory instead?
1. Update business requirements: cost, price, by group, drug, prescriber
2. Identify source data: claims file
3. Update fact/dimension model: claims, group, drug, prescriber, calendar
4. Map source to new DW tables
5. Build new ETL, update old ETL
6. Update report design
7. Map DW to new report fields
8. Update and test reports
Repeat cycle
Traditional Methodology
6. Agile Data Analytics
1. Anchor on user about what they’re trying to accomplish: demonstrate the value of our
program
2. Identify source data: claims file
3. Load data into database
4. Run some simple queries to establish baseline for testing
5. Identify critical data elements: cost, price, by group
6. Create logic to filter, group, and summarize data
7. Run scripted testing routine
8. Run correlation analysis
6
Nothing shows up, so user SME
recommends grouping
Specialty on the claims is pretty
sketchy, can you use this
directory instead?
Repeat cycle
7. Faux Agile: Work happens in “iterations,” but…
• As a result:
• Delivering new features takes a
longer than expected time
• Team members are underutilized
or project-switching
• User is unhappy with resulting
product and timeline
• Team members are frustrated
with the appearance that
“requirements aren’t nailed down,
so we must not know what we’re
doing” and “everything keeps
changing”
7
• Focus has been on the technical
solution, not the value
• Each iteration requires:
• Excessive documentation
• Adding new code
• Updating old unrelated code
• Coordination of several teams
• Manual retesting and validation
Agile Data Analytics Methodology
8. Integrated Team
8
Agile Data Analytics Methodology
SME Team Member
Anchor on goal
Identify source data
Load data
Run some queries
Identify CDE
Design solution logic
Validate with automated tests
Run analysis
9. Creating Effective Agile Data Analytics
• Focus has been on the value,
but keep an eye on the future
as you make design
decisions.
• Documentation – just enough
• Requirements – collaborative
rather than declared
• Architecture – simple and
adaptable
• As a result:
• Faster turn-around time
• Collaborate in real-time on
understanding data
• Making changes is faster
• The “expect change” attitude
means that adding a new
source is just another day. No
big deal.
• Greater morale for the team
9
Agile Data Analytics Methodology
10. Agile Data Analytics
10
Key Take-aways
• Iterative ≠ Agile
• Agile assumes customer participation
• Agile may require a different architecture
• Agile has processes that support it on a daily basis
11. References
• Agile Analytics, Ken Collier, 2011
https://www.amazon.com/Agile-Analytics-Value-Driven-Intelligence-Warehousing/dp/032150481X
• AgileData.org, Scott Ambler
• Agile Data Warehouse Design, Lawrence Corr, 2011
https://www.amazon.com/Agile-Data-Warehouse-Design-
Collaborative/dp/0956817203/ref=pd_lpo_sbs_14_t_0?_encoding=UTF8&psc=1&refRID=ZANFME4BP
FGADKAXNH1Y
11