�ݺ�ߣ

WhatisBigData
 From the beginning of human civilization
until 2003, entire world generated 5
Exabyte of data.
 In 2004, US alone produced 5 Exabyte of
data every two days and the rate of
growth is accelerating in a rapid pace.
 1 Exabyte = 1 Million Terabyte

What’s Big Data?
IT industry defines Big Data using 4
Vs.Volume: amount of
data
Velocity:
speed of
data
arrival
Variety: text,
image, video
Veracity:
Trustworthiness

WhatisBigData
 Volume – Peta Byte, Exa Byte etc.
 Variety – Structured, Unstructured –
video, twitter trends, free form text
 Velocity- streaming data arriving in real-
time
 Veracity – trustworthiness of data
removing biases, noise, abnormality

Initial Challenges of the Big Data
 Prohibitively expensive hardware –
Computing, Networking and Storage
 Small pool of Big Data experts
 Lack of awareness about benefits of data
collected from different sources
 Analytics to process real-time data in
milli-seconds

Tools for Big Data
 Cluster of distributed computing nodes
using commodity hardware
 Map-reduce framework to run parallel
computation on Hadoop Cluster
 No SQL databases – Columnar database
using Key-value storage for very fast
data retrieval

Tools For Analyzing Big Data
Hadoop
 Batch Processing Framework
uses distributed cluster
running on low cost multi core
computers.

Splunk
 Real time analytics software to
process streaming data from
millions of sensors.

 Map Reduce Programming Model
that distributes small chunks of
data across thousand of nodes for
parallel processing and
combines the output from each
node to solve the big data
problem

Big Data Impact at all
Industries
Health Care
 MIT Technology Review report on data-driven health
care using “The New Medical Data Eco System”
 Analytics and predictive modeling of medical data
captured from many sources – insurance claims, public
health data, mobile health data and electronic medical
records to provide personalized patient care and help
doctors quickly decide best treatments

Insurance Claims Data
 Trends in drug and treatment
usage

Environmental Data
 “Sensors can pick up
behavioral Information
 Ex: Mapping, Location, and
Weather Data.”

Genomic Data
“Less expensive genome
sequencing offers insight into
the role genetics may play”

Public Health Data
 “Insight into community health
patterns from federal and state
data.”

Mobile Health Data
 “100-000 plus mobile health
apps, plus wearable devices
that measures activity bodily
function, offer a constant read on
patients.”

Electronic Medical Records
 Digital records include lab and
test results, drug prescription
and physician’s reports.
 These all records create
Family Health History

Outcome
 “Analytic algorithms and predictive
modeling mine the layers of data for
patterns and insight” (MIT Technology
Review)

Outcome
 Patients
“More precise and personalized diagnosis
and care based on a holistic view may
become possible.”
 Doctors
“Decision-support tools could help quickly
evaluate the best treatments”

Outcome
 Researchers
“Detailed information from many
patients, along with other data, could
lead to new insights into disease
and treatment.”

Use Cases Across all Industries
 Recommendation Engine
 Customer Sentiment Analysis
 Marketing Campaign Analysis
 Social Network Analysis
 Fraud Detection
 Risk Analysis

Retail
 Macy’s Inc.
Optimizing pricing of 73 million
items based on real-time market
data.

Retail
 Wal-Mart
1. Display search results based on
semantics of search items predict
customer behavior
2. Customer behavior prediction
3. Supply chain management analysis
of millions of point-of sales data in
real – time.

Future Trends
Health Care
 Important tool for cost reduction
 Adoption of EMR- Electronic Medical Record
for better patient treatment plan
 Agriculture
 Real-time tracking of farm machinery

FutureTrends
 Internet protocol becoming standard in
electricity grid, oil industry etc..
 IP v6 with 128 bit address will theoretically allow
trillion of trillion sensors to connect previously
unconnected places, people and things
 Digitization of massive data currently stored in
non-digital form

Citation
 Big Data@Work, Thomas H.Davenport
 MIT Technology Review
 Harness the Power Of Big Data, Paul Zikopolos,
Dirk Deroos, Krishna Parasuraman, Thomas
Deutsch, David Corrigan, James Giles
 The Human Face of Big Data, Rick Smolan and
Jennifer Erwitt

�ݺ�ߣ

Big data

More Related Content

Big data

Editor's Notes