Banking on Analytics

1
Banking on Analytics

• Dr A S Ramasastri
• Director, IDRBT

2
A few questions . . .

• What is the impact on sales and profit by a new
  product / service introduced by you?
• What is the general opinion in the market on a
  product / service introduced by you?
• Who is the ideal customer to whom you can make a
  personal offer of the product / service? Is the
  particular customer worthy of the offer?
• What would happen if you make a few changes to
  the product / service?
• Are there any demography-based linkages among
  products, services, defaults and frauds?

3
. . . and approaches to answers

• Reports from data warehouse / data mart thru OLAP
  tools Business Intelligence
• Opinion Mining on Social Networks Descriptive
  Analytics
• Finding potential customer and her value based on
  past behavior Predictive Analytics
• Assessing the impact of an action on a result
  Prescriptive Analytics
• Exploring huge volume of data for discovering
  hidden patterns Data Mining

4
The need of the hour

• Relevant Quality Data
• Qualified Data Scientists
• Coordinated Efforts by Concerned Companies
• Focused Applied Research by Institutions with
  support from companies and bodies
• In case of banks, IDRBT has initiated the process
  with the support from stakeholders

5
IDRBT

• A unique institute established by Reserve Bank of
  India for development and research in banking
  technology
• Works closely with Reserve Bank of India, banks
  and academicians on important areas of
  application of technology in banks information
  security, payment systems, networks, cloud
  computing and analytics

6
Analytics Center at IDRBT

• Lab exclusively for analytics has been set up at
  IDRBT a few years back
• Banks have training programs and experiments
  conducted at IDRBT lab both at individual bank
  level and bank group level
• The areas of focus are generally CRM, risk
  management and fraud analytics
• Dedicated faculty and research scholars

7
CRM Products and Services

• Customer Retention customer behavior prior to
  attrition, model to retain the customers
• Targeted Marketing identify buying patterns,
  finding associations among customer demographic
  customers, predicting response to various types
  of campaigns
• Credit Card identifying loyal customers,
  predicting customers likely to change their
  affiliation, determine card using behavior,
  selecting appropriate product / service

8
Assessment Credit and Portfolio

• Credit Appraisal based on the data on the
  current customers, develop classes of
  risk-worthiness and classify a new borrower into
  one of the classes
• Portfolio Management identifying trading rules
  from historical data, selecting financial assets
  to be included in the portfolio, assessing impact
  of market changes on portfolio optimizing
  portfolio performance

9
Prediction Defaults and Frauds

• Housing Loan Prepayment Prediction
• Mortgage Loan Delinquency Prediction
• Uncovering hidden correlations between customer
  characteristics and behavior
• Detecting Patterns of Frauds Credit Card, ATM,
  Internet Banking Frauds
• Real Time Alerts on Online Frauds

10
Tools for Analytics / Data Mining

• Classification
• Clustering
• Correlation
• Regression
• Association Rule Learning
• Pattern Recognition
• Deviation Detection
• Artificial Neural Networks

11
Some Open Source Software

• R
• RapidMiner
• OpenNN
• Orange
• Apache Mohout
• KNIME
• Weka

12
Further References

• Google !!!
• After all
• Google MUST be using
• several techniques
• to analyze such large volumes of web data

13
Thanks

• asramasastri_at_idrbt.ac.in