Krea University

Executive Education

Big Data and Advanced Analytics

Nov 2022
Objective
• Understand What is Big Data or Data Lake
• What is the difference between Data Warehouse and Data Lake
• Learn how banks can leverage Data Lake
• Appreciate various Banking Advanced Analytics Usecases based on
Data Lake
Technology Play in Banking
Technology is playing a significant role in Banking Industry for Several
decades now

Existing Technology Emerging Technology

• Omnichannel Services • Blockchain

• Mobile and Internet Banking • Automation
• AI / Machine Learning
• Core Banking Solution
• Data Lake & Advanced Analytics
• Data Warehouse and Analytics
• Cloud
• ERP • ISV Based Transformation
• IoT
Data Lake and Advanced Analytics
Database, Data Mart, Data WH and Data Lake

Database & RDBMS - A Data Mart - A data mart Data Warehouse - A

database is an organised is a subset of a data data warehouse is a Data Lake - A data lake is a
collection of structured warehouse focused on a central repository of centralized repository
information, or data, particular line of information that can be designed to store, process,
typically stored business, department, analyzed to make more and secure large amounts of
electronically in a computer or subject area. Data
system. A Relational informed decisions. structured, semistructured,
marts contain and unstructured data. It
Database RDBMS is a type of Data flows into a data
repositories of
Database that stores data in summarized data warehouse from can store data in its native
a row-based table structure collected for analysis on transactional systems, format and process any
which connects related data a specific section or unit relational databases, variety of it, ignoring size
elements. Examples of the within a Bank like and other sources, limits. Data Lake holds big
most popular RDBMS are Lending, Treasury,
MYSQL, Oracle, IBM DB2, typically on a regular data from many sources in a
Credit Cards, Mortgages cadence raw, granular format.
and Microsoft SQL Server etc
database
What is Database, Data WH and Data Lake

Data Lake Key Advantages

Structured and Unstructured Data
More Granular data
Leverage by AI / ML
Faster Analysis of Data
Big Data Significance
12 terabytes 5 million Derive cohesive, actionable
of Tweets trade events insights from disparate sources of
created daily per second
varying volumes, velocity, variety
and veracity.
Volume Velocity

Variety Veracity

1000’s Only 1 in 3
Of video feeds from Decision
surveillance cameras makers trust
their
information

“Data is the new oil.”

Clive Humby

7
What is a Data Lake ?
A repository of all the organization’s structured
and unstructured data based on big data
technology example Hadoop platform

It is built on inexpensive hardware with design

principles of

• Ingest everything
• Dive in anywhere
• Flexible provisioning

Ability to store and process huge amounts of

data which otherwise is not possible using
traditional databases

Easily scalable by adding additional nodes to

handle increased volumes

Sandbox for exploratory analysis

8
Why data Lake
Numerous, poorly rationalized
Current environment in most Banks sources lead to conflicting data
downstream

For new data, you can

 Multiple redundant data feeds and databases. wait in line or do-it-yourself
 Data for analytics purposes is typically not readily Duplicative, slow and poorly aligned
available except through traditional approaches. to business consumption
 The resultant costs and delays have lead to the creation of
multiple ungoverned shadow databases. Limited by what you get,
hard to rationalize & explain

Build Data
Lake

Organized and focused

extraction of source data sets
Future environment
Raw, lightly-curated and heavily-curated
access depending on user skill/need  Reduce costs
Reliable, flexible and part of an agile data
 simplifying the environment,
process
 data much more accessible.
Rationalized marts using curated data that
support what users need
Why Data Lake ?
Traditional DW Architectures The Data Lake Advantage

• Brittle architectures with rigid schemas • Cost effective (approximately savings of

• Suitable for high value analytic data sets 30-50X on storage)
• Inability to process unstructured data • Can handle both scale and variety - semi-
structured and unstructured data
• Expensive licenses; price per Terabyte for
an MPP-type appliance could go upwards • Flexible, minimalistic schema
of $15 K • Reliable Compute and Dependable Storage
• Large amounts of dormant data in the • Complementary to existing strategies
warehouse • Value and Returns to market
• Ineffective archival and storage strategies

Core
transactions

10
Data Lake Capabilities
Reduce Data Redundancy, Improved / Streamlined Data
Improved Information & Analytics
Movement & Tools Governance & Demand
Management
Quality Efficiency Governance

Single Source of Deep Information Assets Enhanced BI & KPI

Integrated Tool Set
Reporting
Speed Access Toolset

Benefits

Volume Early Structured/ Schema on

Storage Single Source
Analysis Insight Unstructured Read

Provides Supports integration It’s schema on read

Capable of analyzing Provides capability Enables single
capability to of structured and approach of raw file
huge data volumes by to store enormous sourcing of data
apply analytics on unstructured data. formats; integration
pushdown of analytic volumes of data on based on “Write
data close to Provides of new data sources
workload to Hadoop a commodity once read many
source and environment for is quicker/cheaper
clusters hardware times” principle
leverage early performing new /
insight ad-hoc analytics

Reduce costs , simplified environment and much more accessible data

Advanced Analytics Tools ? Types of Analytics
• Advanced Analytics is the • Descriptive Analytics is the simplest type of analytics and the
autonomous or semi- foundation on which the other types are built on. It allows you to
autonomous examination What pull trends from raw data and succinctly describe what happened or
of data or content using is currently happening. Descriptive analytics answers the question,
sophisticated techniques Microsoft Excel
“What happened?”
and tools, typically beyond Python• Diagnostic Analytics includes comparing coexisting trends or
those of traditional
business intelligence (BI), R movement, uncovering correlations between variables, and
Why determining causal relationships where possible. Diagnostic Analytics
to discover deeper Jupyter Notebook
addresses the next logical question, “Why did this happen?”
insights, make predictions, Apache• Predictive
Spark Analytics can make informed predictions about what the
or generate
recommendations. SAS future could hold by analysing historical data in tandem with industry
trends. Predictive Analytics is used to make predictions about future
• Unlike traditional Microsoft
What Power BI
trends or events and answers the question, “What might happen in
analytics, advanced Tableau the future?”
might
analytics can cope with • Prescriptive analytics takes into account all possible factors in a
and extract meaning from scenario and suggests actionable takeaways. This type of analytics
complex data, https://online.hbs.edu/blog/post/prescriptive-analytics
can be especially useful when making data-driven decisions.
unstructured data, and What Prescriptive analytics answers the question, “What should we do
partial or incomplete data To do next?”
Banks can reap in significant benefits by implementing
potential use cases of data lake solution
Increased customer
retention and Increased customer
Increased sales profitability satisfaction
Optimize Offers and Customer Insight and Contact Center Service
Cross Sell Profitability Optimization
How can I deliver more
timely, relevant offers and How can I anticipate customer How can I better understand
improve response rates? activities and better understand customer issues and resolve them
needs? more efficiently?

Reduced operational
risk Reduced credit risk Reduced fraud
Enterprise Operational Risk Fraud Detection and
Management Credit Risk Management Mitigation

How can I monitor internal systems How can I better manage credit How can I better predict, detect and
activity for outages and risks? worthiness and changes in financial investigate fraud?
stability?

Increased profitability

Asset Optimization

How can I Improve trading decisions,

portfolio compositions and
valuations?
How Data Lake solution powers change and drives value
Understand
client across
products to
know more

Consumer
Banking

14
How Bank treats customer today !
Today Tom is treated like any other customer in her segment…

…but Tom is an individual Bank: “Hi

<NAME>!
Can we
interest you in
a credit
card?”

Tom: “Oh,
look! More
junk mail from
the bank…”

15
By using only limited segmentation, Bank treats Tom like
anyone else……and bases its actions by her segment

Model Scoring
Tom holds a
mortgage and Cash Management Acct.
a savings
account with
us Set meeting with Personal
Banker for a
Review

Equity Bank Line /

Secured Line-of-Credit
Tom’s current
credit score &
Preferred Gold Credit
profitability Card
qualifies him
for a preferred
rate

16
Information helps bank understand how Tom is
different, is bank using it?
Tom holds a
mortgage and
a savings Last week Tom Tom has also
account with asked the Call posted property
us Center about photos to
loan processing Facebook asking
times friends to vote

This week, he
checked
Tom’s current mortgage rates
credit score & on the bank web
profitability site three times And today he’s
qualifies him tweeted a link
for a preferred to an article
rate about buying a
second home

17
By using all the information bank can make its service
unique to Tom
Last week Tom has also Tom’s current And today he’s
This week, he
Tom asked the posted property credit score & tweeted a link to
checked mortgage
Call Center about photos to profitability an article about
rates on the Web
loan processing Facebook asking qualifies his for a buying a second
Site three times
times friends to vote preferred rate home

Tom holds a
mortgage and
a savings
account with
us Model Scoring

Cash Management Acct.

Preferred Gold Credit Card

Equity Bank Line /

Secured Line-of-Credit

Mortgage with special rate

discount of 25 basis points

18
Attrition Path Analysis
“Alarming number of When clients leave
accounts attrite per the Bank, we lose
week, but we lack deposits, and also
complete, granular data potential value of up-
to understand why...” sell/cross-sell…”

 Data granularity/frequency:
 Lack of daily client transaction data means business may not know a client is “gone”
until too late to act
Problem

 Lack of daily deposits data for a more complete picture of product utilization, rates,
and fees
 Fragmented data sources makes it difficult to detect changes in clients’ circumstances
that could indicate new financial needs or impending churn

 Analyze behaviors of accountholders who have left the Bank to detect patterns and
Solution
Data
Lake

drivers of checking account attrition, then intervene with the “right” course of action.
 Patterns / drivers of attrition, to generate early warning signals for intervention.
Customer Channel Productivity
Our most expensive
channel, but we don’t
know which client
interactions/messages
result in a sale…”

 Data granularity/frequency:
 Lack of daily client-level transaction and interaction data to track/model behaviors
Problem

and outcomes
 Inability to capture/analyze semi- or unstructured data such as Client Needs Met
inputs

 Understand which client interactions drive positive outcomes (e.g., result in a

Data Lake

sale/high client satisfaction) based on high-performing Team Member/Branch

Solution

activities or techniques
 Leverage interactions that drive positive results to improve under-performing
resources/locations.
Attracting HNW Clients
Combined, HNW clients hold
Trillions in assets someplace
other than my Bank…how do
I utilize their money
better..?

 Data granularity/frequency: Need daily client-level transactions to understand

product usage and client investment activities/preferences
Problem

 Fragmented data sources results in lack of visibility to flow of funds between

deposits and investments accounts to determine when/how money is moving
and formulate an effective response.

 Access and visibility to integrated client-level daily deposits and investments

Data Lake

data, for visibility to product utilization and flow of funds between

Solution

products/accounts for Affluent and High Net Worth client segments.

 Understand client deposits / investments flow of funds to capture incremental
revenue.
 GAIN A 360° VIEW OF CUSTOMERS

Sharpen your Unstructured Structured

view of each information information
customer by Geolocation
leveraging more Customer
Web click demographics
data—and more streams
types of data
from more Account
Social media posts information
sources—than
ever before.
Correspondence Credit data

Contact center notes Transaction

and chats details
Channel usage
 USE GREATER VISIBILITY:

PERSONALIZE SALES AND MARKETING

Descriptive data Attitudinal data Behavioral data Interaction data

Generalized offers aimed at Segment-of-one
groups marketing and sales

Leverage large volumes of Empower the CMO with insight into Speed the capture and analysis
BOTTOM-LINE
multistructured data to customer channel preferences to of customer data to enhance
BENEFITS inform marketing decisions improve the efficiency and impact of customer service and sales
and better target customer marketing campaigns. interactions.
offers.
 SPEED AND ENRICH ANALYSIS TO ANTICIPATE CUSTOMER’S UPCOMING NEEDS

Use predictive
Descriptive data Attitudinal data Behavioral data Interaction data
analytics to
determine “next
best action” with
greater certainty 39-year-old Posted a link Looked on the Made two calls in
professional to an article last bank’s website for the last week to the
and speed. night about current mortgage contact center about
Been in
buying a rates home loan
her current
second home processing times
job more than 10
years “Liked” three
vacation
properties

NEW INSIGHT:
Anne is shopping for a vacation home.
 IDENTIFY NEXT BEST ACTION:
PROVIDE AN OFFER THAT ANNE IS MOST LIKELY TO ACCEPT

Text a link to a customer mortgage calculator to help the

customer make an informed decision.

Offer a preferred rate mortgage with reduced

closing fees.

BOTTOM-LINE Enrich your predictive Empower sellers Use sales and marketing
BENEFITS models with more robust with faster customer insights to enhance customer
analytics to help the CMO insights to uncover new satisfaction and increase
pinpoint offers customers cross-sell and up-sell share
are most likely to accept. opportunities sooner. of wallet and loyalty
over time.
Top Big Data Financial Services Use Cases

https://www.safegraph.com/blog/top-big-data-use-cases-financial-services
Thank You