Creating a Smarter Bank - FST · that conventional data modelling is just not done. Kiwibank...
Transcript of Creating a Smarter Bank - FST · that conventional data modelling is just not done. Kiwibank...
Creating a Smarter Bank
Bohdan Szymanik, Kiwibank
http://bohdanszymanik.blogspot.com
Objective: get more productive!
Execution
Decision Making
Intelligence
Information
Communication Challenge
2001 2002 2003 2004 2005 2006 2007 2008 2009
Infrastructure Staff
How About Kiwibank?
Conventional BI Isn’t Enough
“Focus on Data Management”
“Focus on Financials”
“It has to be right!”
“Focus on making it easy”
Where’s the data?!
What do we do?
We look for analogues!
Lessons from Systems Management
•Big data
• Diverse data
• Context
• Health indications
Sampling
Hierarchical ModelsBespoke Analysis
Monitors and Trends
The volume and diversity is so great that conventional data modelling
is just not done.
Kiwibank Experience: System
Center
Scheduler Component
Server AScheduler Component
Server B
System Center = A Data Source
What have we learnt?
• Sample data
• Don’t bother with relational models
• Model context can be very simple
– Graphical
– Descriptive
• Make data available true to source
• Expect end user analysis
Empowering the analyst!
Make data available and provide end user toolsAnalysts create local models and local queries
Data Analyst
+ Model Analyst
+ Query Analyst
= A Tough Combination?
Maybe Not…
Localised model creationIn memory databasesContent management
Empowered Analyst
Customer>1,000,000
records
Investment Accounts
<1,000,000 records
Who do we get our money from!?
HP Mini2 GB RAMIntel Atom
Sharepoint2010
My new marketing campaign!
$ as a function of (Surname first letterSurname length )
A
C
E
G
I
K
M
O
Q
S
U
W
Y
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Web 2 / Enterprise 2 = new
data sources!
Social AnalyticsDS 4.3CB 3.0BM 2.0JC 1.7DO 1.4KT 1.0BS 1.0NC .9JW 0.9AH 0.9CB 0.9SK 0.8TB 0.7DG 0.6NL 0.5BN 0.5
How?• Code – Team System• Documents, Wiki, Blog – Sharepoint
Example:Systems of more importance have had more change over time. Correlate change to individuals and account for shared knowledge to identify where people have worked on many small projects, and therefore present significant key man risk.
I keep saying the sexy job in the next
ten years will be statisticians.
Hal Varian, Google’s Chief Economist, The
McKinsey Quarterly, January 2009
http://www.nytimes.com/2009/01/07/technology/business-computing/07program.html?_r=1
http://www.forbes.com/forbes/2010/0524/opinions-software-norman-nie-spss-ideas-opinions.html
Summary
1. Don’t push non-financial data through conventional BI stacks
2. Empower analysts
– Make information available
– Use minimal models
– Don’t be scared of desktop query
– Don’t be scared of code