BIG DATA

Name: KARISHMA N. PARAB GAONKAR

ASSISTANT PROFESSOR
COMPUTER ENGINEERING DEPARTMENT
GOA COLLEGE OF ENGINEERING

1
 Content

 Data , Information , Knowledge

 Example
 Big Data
 Examples
 Types of Big Data
 Characteristics of Big Data
 Big Data Analytics

2
 DATA,INFORMATION ,KNOWLEDGE

 Data : The quantities , characters or symbols on which

operation are performed , which may be stored and
transmitted over a medium.
 Example : weights, price, cost, number, name etc.

 Information: It is a data that has been converted into useful

or intelligible form.
 Example: Report Card, Merit List, Documents etc.

 Knowledge : Ability of person to use his information for a

specific purpose.
3
 Example

 “386” is a data

 “Your marks are 386 out off 400” is information

 “It is a result of your hard work” is knowledge

4
 Big Data
 Its also a data with huge size.

 It is a term used to describe data which is huge in volume and

yet grows exponentially over a period of time.

 Not suitable for traditional data management tools.

 The challenges include capture, duration, storage, search,

sharing, transfer, analysis, and visualization.

5
 Examples

 Stock market

 Social Media

 Sensors

 Web

6
 Types of Big Data

 Structured

 Unstructured

 Semi-Structured

7
 Structured Data
 Data that can be Stored , Accessed and Processed using fixed
format.

 It contains rows and columns.

 Easy to derive values.

 Example : Relational databases

8
 Unstructured Data
 Any data with unknown form or structured is classified as
unstructured data.

 Size is huge , involved multiple challenges to process the

data.

 It contain heterogeneous data : text files , videos , images etc.

 Example : Output of google search

9
 Semi Structured Data

 It contain both forms of data.

 Example : XML file , JSON etc

10
 Characteristics Of Big Data

 Volume

 Variety

 Velocity

 Variability

11
 Volume

Exponential increase in collected/generated data

12
 4.6 billion
30 billion RFID tags camera
today phones
12+ TBs (1.3B in 2005) world wide
of tweet data
every day

100s of
millions of
GPS enabled
data every day
? TBs of

devices sold
annually

25+ TBs of
2+ billion
log data
every day people on
the Web by
end 2011
76 million smart meters in
2009…
200M by 2014
13
 Variety

Social Banking
Finance
Media

Our
Gaming
Customer Known
History

Entertain Purchase

14
 Velocity

Mobile devices
(tracking all objects all the time)

Social media and networks Scientific instruments

(all of us are generating data) (collecting all sorts of data)

Sensor technology and networks

(measuring all kinds of data)

15
 Some Make it 4V’s

16
 Growth of Big Data

17
 Big Data Analytics
 It is complete process of collecting, gathering, organizing and
huge sets of data(Big Data) to identify pattern and to extract
other useful information to make a decision.

 Big Data analytics can be used in various sectors like media,

education, healthcare, various government and non
government sectors etc.

18
 example

 Uber- driver, vehicle, location etc. fairs, location etc can be

predicted.

 Banking – fraud detection

19
 THANK YOU

20
 content

 Structured data vs. Unstructured data

 Big Data vs Conventional Data

 Techniques used by big data

 Importance of Big Data Analytics

 Sources of Big Data

 Challenges of Big Data

 How big data differs

 Web data

1 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Structured data vs. Unstructured data

Structured Data Unstructured Data

Predefined data models No predefined data model
Usually text only May be text,images,sound,videos or other
formats.

Easy to search Difficult to search

Resides in relational databases Resides in non relational databases

Generated by humans or machines in airline Generated by humans and machines through word
reservation systems, inventory control systems, processing, email clients, tools for viewing or
ERP systems etc. editing media.

2 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Big Data vs Conventional Data
Big Data Conventional Data
Huge data sets. Data set size in control.
Unstructured data such as text, video, Normally structured data such as numbers
and audio. and categories, but it can take other forms
as well.
Hard-to-perform queries and analysis. Relatively easy-to-perform queries and
analysis.
Needs a new methodology for analysis Data analysis can be achieved by using
conventional methods.
Need tools such as Hadoop, Hive, Tools such as SQL, SAS, R, and Excel
Hbase, Pig, Sqoop, and so on. alone may be sufficient.
Raw transactional data. The aggregated or sampled or filtered data.

Generated by big financial institutions, Generated by small enterprises and small

Facebook, Google, Amazon, eBay, banks.
Walmart, and so on.

3 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Techniques used by big data
 Distributed storage and processing

 Non-Relational database

 Streams and Complex Event processing

 Data processing

 In-Memory Processing

 Reporting Layer(visualization)

4 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Importance of Big Data Analytics

 Cost reduction.

 Faster, better decision making.

 New products and services.

5 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Sources of Big Data

 Social Networks(facebook,twitter etc)

 Traditional Business Systems(banking, e-commerce etc)

 Internet of things(satellites images, sensors etc)

 etc

6 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Challenges of Big Data
 Meeting the need for speed :
 Not only to find relevant data but to find quickly.
 Granularity of data
 Solution: hardware with increase memory and parallel processing.
 Grid computing approach

 Understanding the data

 Visualization
 Customer using a set of product and understand what your trying to
infer from this data.
 Solution : proper domain expertise

7 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Challenges of Big Data
 Addressing data quality
 accuracy of data for decision making
 Solution : companies should have information management process to
ensure data is clean.
 Use of proactive method(solution before they appear)

 Displaying meaningful results (clustering)

 Plotting points is difficult(many datapoints)
 Solution: cluster data for better visualization

8 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Challenges of Big Data
 Dealing with outliers
 The graphical representations of data made possible by visualization
can communicate trends and outliers much faster than tables
containing numbers and text.
 Solution :remove the outliers

9 Prof. Karishma Parab ,Computer Engineering dept,GEC

 How big data differs
 Automatically generated by machine

 Entirely new source of data

 Not a user friendly

 Need to focus on important part.

10 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web Data
 Data collected over the internet specific to a corporate or to
an individual.

 It’s a versatile data and its used for predictive analysis, that’s y
it is considered as a most popular big data.

 360 view of particular item.

11 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web Data

 Earlier company has started with basic recency, frequency,and

monetary value(RFM)metrics attached to customers.

 But now companies have newly evolving big data sources such
as data from web browsers, mobile applications, social media
sites etc

12 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web Data
 We might have missing data so improvement will be to know
everything about the customer does as they go through the process
of your organization.(new level of understanding and interaction)

 Web data provides new level of information, i.e. it provides

factual information on customer preferences, future intentions,
and motivations that are virtually impossible to get from other
sources

 Privacy :remove privacy concern by use an arbitrary identification

number for each customer,

13 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web Data
 What Data should be Collected?
Purchases
Requesting help
Product views
Forwarding a link
Shopping basket additions
Posting a comment
Watching a video
Registering for a webinar
Accessing a download
Executing a search
Reading / writing a review

14 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web Data help organization
 Shopping Behaviour (search term, search engine, reference
sites)
 identify products that are of interest to a customer before they make
a purchase.

 Research Behaviour (make easy to fine their favourite option)

 Customers who goes for specification ,customers who go for photos

 Feedback Behaviour

15 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Web data in action
 Attrition Modelling : Those customers who are most at risk
of cancelling their accounts, so that action can be taken
proactively to prevent them from doing so.

 Response Modeling: Many models are created to help predict

the choice a customer.

 Customer Segmentation: segment customers based solely

upon their typical browsing patterns.

16 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Evolution of Analytic Scalability

1 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Analytic Scalability

 New level of scalability.

 As the amount of data organizations process continues to

increase, the same old methods for handling data just won’t
work anymore.

 Organization has to update their technologies to provide a

higher level of scalability.

2 Prof. Karishma Parab ,Computer Engineering dept,GEC

 The Convergence of the Analytic
and Data Environment
 So earlier Data was not on one place and tools required to run
the analysis was not able to find data.
 Analytic professionals had to pull all their data together into a
separate analytics environment to do analysis.

 Analyst did Merging of data.

 Analysts do what is called “data preparation.”(pull and merge)

 In data warehousing world this process is called “extract,
transform, and load (ETL).”

3 Prof. Karishma Parab ,Computer Engineering dept,GEC

  Earlier they used mainframe, then they came up with RDB
(scalability).

 Data Marts: Databases were built for each specific purpose or

team, and relational databases were spread all over an
organization.

 Enterprise Data Warehouse (EDW):Combining the various

database systems into one big system.

4 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Traditional Analytic Architecture
Old way of doing things.

5 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Modern In-Database Architecture

•Processing stays in database.

•User machine has to just submit the request

•This is the concept of In Database Analytics

6 Prof. Karishma Parab ,Computer Engineering dept,GEC

 MASSIVELY PARALLEL PROCESSING
SYSTEMS
 It is a mechanism for storing and analyzing large amounts of
data.

 An MPP database spreads data out into independent pieces

managed by independent storage and central processing unit
(CPU) resources.

 It removes the constraints of having one central server with

only a single set CPU and disk to manage it.

7 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Massively Parallel Processing
System Data Storage

8 Prof. Karishma Parab ,Computer Engineering dept,GEC

 An MPP database breaks the data into independent chunks with
independent disk and CPU

100-gigabyte 100-gigabyte 100-gigabyte 100-gigabyte 100-gigabyte

chunks chunks chunks chunks chunks
One-terabyte
table 100-gigabyte 100-gigabyte 100-gigabyte 100-gigabyte 100-gigabyte
chunks chunks chunks chunks chunks

A Traditional database will

query a one-terabyte table 10 simultaneous 100-gigabyte queries
one row at time

9 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Massively Parallel Processing
System Data Storage

 This allows much faster query execution, since many

independent smaller queries are running simultaneously
instead of just one big query.

 It gets a little more complicated in cases where data must be

moved, but MPP handles that in a very fast way.

10 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Traditional Query versus an MPP
Query

11 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Advantages of MPP

 Data is stored at multiple location(redundance),so incase of

failure recovery is faster.

 Uses resource management tools to manage the CPU and

disk space.

 query optimizers to make sure queries are being optimally

executed.

12 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Using MPP Systems for Data
Preparation and Scoring
 4 Primary ways

1.SQL Push Down

2.User-Defined Functions (UDFS)

3.Embedded Processes

4.Predictive Modelling Markup Language (PMML) Scoring

13 Prof. Karishma Parab ,Computer Engineering dept,GEC

 SQL Push Down
 SQL is the native language of an MPP system and it’s efficient
for a wide range of requirements.

 Many Data Preparation task can be translated into SQL and

push it down to the database.

 Analytic tools will often translate the logic from a model into
SQL for the user or user can code an SQL script.

 So the data preparation or scoring processes end up being

executed purely with SQL.

14 Prof. Karishma Parab ,Computer Engineering dept,GEC

 User-Defined Functions
 It allows user to define logic that can be executed in same manner
as SQL function does.

 It involves compiling code into new database functions that can be

called from an SQL query.

 User-defined functions are coded in languages like C++ or Java.

 analytic professionals don’t know how to program, so we can use

analytic tools for generating appropriate UDF and load into db.

15 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Embedded Processes
 Its an analytic tool’s engine actually running on the database
itself.

 An embedded process is therefore capable of running

programs directly inside the database.

 There is no translation required.

 This method requires the least amount of changes to the

code.

16 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Predictive Modeling Markup
Language
 Predictive modeling markup language (PMML) is a way to
pass model results from one tool to another.

 The type of information included in a PMML feed includes a

model type, the variable names and formats, and the
parameter values.

 Drawback : The exact same variables in the exact same

format must be available in the system where the PMML is
being deployed.

17 Prof. Karishma Parab ,Computer Engineering dept,GEC

 Content
 Cloud Computing – Public Cloud, Private Cloud

 Grid Computing

 Analytical Process

 Analytic Sandbox, Benefits

1 Prof. Karishma Parab ,Computer Engineering dept,GE

 Cloud Computing
 It is the delivery of different services through the internet.

 These resources includes tools and applications like data

storage, servers, databases, networking and software.

 Instead of keeping files on proprietary storage you store it on a

remote database.

 So user is not required to be at specific place to access the data

as data is stored remotely.

2 Prof. Karishma Parab ,Computer Engineering dept,GE

 Criteria for cloud environment
 3 criteria

 Enterprises incur no infrastructure or capital costs, only

operational costs.(pay-per-use basis).

 Capacity can be scaled up or down dynamically, and

immediately.

 The underlying hardware can be anywhere

geographically.(multi-tenancy mode)

3 Prof. Karishma Parab ,Computer Engineering dept,GE

 Characteristics of Cloud
environment
 5 characteristics of a cloud environment.
1. On-demand self-service

2. Broad network access

3. Resource pooling

4. Rapid elasticity

5. Measured service

4 Prof. Karishma Parab ,Computer Engineering dept,GE

 Types of cloud environments
Two types of cloud environment

 Public clouds

 Private clouds

5 Prof. Karishma Parab ,Computer Engineering dept,GE

 Public Clouds

 Users are basically loading their data onto a host system

 Users are then allocated resources to use this data.

 So users get charged accordingly.

6 Prof. Karishma Parab ,Computer Engineering dept,GE

 Advantages of public cloud
 The bandwidth is as needed, as user pay only for what they use.

 No need to buy system of high capacity and then having risk of half of
the capacity unused.

 Simply pay for extra resources if needed for processing.

 Once access is granted user load their data and start analyzing.

 Easy to share data with others regardless of location. Since public

cloud is outside the firewall

7 Prof. Karishma Parab ,Computer Engineering dept,GE

 Disadvantages of Public Cloud
 There are few performance guarantees in a public cloud.(multiple
users can request for same resources)

 Can cause high variability in performance.

 Security of data

 It can get expensive if a cloud isn’t used wisely(bad query)

 If an audited trail of data and where it sits is required, it is not

possible to have that in a public cloud

8 Prof. Karishma Parab ,Computer Engineering dept,GE

 Private Cloud

 Owned exclusively by one organization and typically housed

behind a corporate firewall.

 A private cloud is going to serve the exact same function as a

public cloud, but just for the people or teams within a given
organization.

9 Prof. Karishma Parab ,Computer Engineering dept,GE

 Public Clouds versus Private Clouds

10 Prof. Karishma Parab ,Computer Engineering dept,GE

 Advantages of private cloud

 Organization has complete control over data and system

security.

 Data won’t leave a complete firewall ,so data is at low risk.

11 Prof. Karishma Parab ,Computer Engineering dept,GE

 Disadvantages of private cloud

 It is necessary to purchase and own the entire cloud

infrastructure before allocating it out to users.

12 Prof. Karishma Parab ,Computer Engineering dept,GE

 GRID Computing

 Used for some computations and algorithms which cannot be

converted into SQL or embedded in UDF.

 So you pull out a data into a more traditional analytics environment

and run analytic tools against that data in the traditional way.

 So large servers are utilized to do this task , it leads to expanding the

size and number of servers.

13 Prof. Karishma Parab ,Computer Engineering dept,GE

 Grid Computing
 A grid configuration can help both cost and performance.

 Instead of having a single high-end server, a large number of

lower-cost machines are put in place.

 So jobs are given to individual servers, and are processed

simultaneously

 Each machine may only be able to handle a fraction of the work

of the original server and can potentially handle only one job at
a time.

14 Prof. Karishma Parab ,Computer Engineering dept,GE

 Grid Computing

 Disadvantage - Big jobs won’t run as quickly on the cheaper

machines as on a larger server.

 So the innovation would be to use a high performance analytics

architectures, where the various machines in the grid are aware
of each other and can share information.

15 Prof. Karishma Parab ,Computer Engineering dept,GE

 Analytic Process
 Updating technology for scalability won’t be beneficial if your
using old analytic processes.

 As analytic professionals begin constantly using a database

platform for their work.

 So you provide workspace or sandbox within the database

system for analytic professional.

16 Prof. Karishma Parab ,Computer Engineering dept,GE

 Analytic Sandbox
 Analytic sandbox provides permissions and access to use
enterprise data warehouse or data marts effectively by the
analytic professionals.

 An analytic sandbox provides a set of resources with which in-

depth analysis can be done to answer critical business questions.

 An analytic sandbox is ideal for data exploration, development

of analytical processes, proof of concepts, and prototyping.

17 Prof. Karishma Parab ,Computer Engineering dept,GE

 Sandbox
 A sandbox is basically a small set of users which creates data
within the sandbox which is segregated from the production
database.

 Sandbox users will be allow to load their data which is required

for a project, which may not be a part of enterprise data model.

 Data in the sandbox will have limited life ,i.e. build a data
needed for project and delete it as soon as project is done.

18 Prof. Karishma Parab ,Computer Engineering dept,GE

 Analytic Sandbox Benefits

 Benefits from the view of an analytic professional

 Benefits from the view of IT

19 Prof. Karishma Parab ,Computer Engineering dept,GE

 Benefits from the view of an
analytic professional:
 Independence (no need to ask permissions)

 Flexibility (can use business intelligence, statistical analysis,

visualization tools that they need)

 Efficiency(no need to migrate the data)

 Freedom (no need to focus on administration system, you can just

shift the maintenance task to IT)

 Speed (parallel processing)

20 Prof. Karishma Parab ,Computer Engineering dept,GE

 Benefits from the view of IT
 Centralization

 Streamlining (both development and deployment)

 Simplicity

 Control(production environment is safe if experiment went

wrong)

 Costs (data marts in one central system)

21 Prof. Karishma Parab ,Computer Engineering dept,GE

 Content
 Internal External & Hybrid Sandbox

 Data analytic tool

 Analysis vs Reporting

1 Prof. Karishma Parab ,Computer Engineering dept,GE

 Internal Sandbox
 The sandbox is physically located on the production system.
 The sandbox database itself is not a part of the production
database.
 The sandbox is a separate database container within the system.

2 Prof. Karishma Parab ,Computer Engineering dept,GE

 Advantages
 Easy to setup(use of existing resources and infrastructure)

 Ability to directly join production data with sandbox data.

 Production data and all of the sandbox data are within the
production system so easy to link resources.

 An internal sandbox is very cost-effective since no new

hardware is needed

3 Prof. Karishma Parab ,Computer Engineering dept,GE

 Disadvantages
 There will be an additional load on the existing enterprise data
warehouse or data mart.

 The sandbox will use both space and CPU resources (potentially
a lot of resources).

 Internal sandbox can be constrained by production policies and

procedures.

4 Prof. Karishma Parab ,Computer Engineering dept,GE

 External Sandbox
 For an external sandbox, a physically separate analytic sandbox is
created for testing and development of analytic processes.

5 Prof. Karishma Parab ,Computer Engineering dept,GE

 Advantages

 The sandbox is a standalone environment, dedicated to

advanced analytics development.

 Allows for flexibility in design and usage.

 External sandbox reduces workload ,because only analytic

professionals are using the system,

6 Prof. Karishma Parab ,Computer Engineering dept,GE

 Disadvantage

 Additional cost of the stand-alone system that serves as the

sandbox platform.

 There will be some data movement from the production system

into the sandbox before developing a new analysis.

7 Prof. Karishma Parab ,Computer Engineering dept,GE

 Hybrid Sandbox
 A hybrid sandbox environment is the combination of an internal
sandbox and an external sandbox.

 It allows analytic professionals the flexibility to use the power of the

production system when needed, and also external system.

8 Prof. Karishma Parab ,Computer Engineering dept,GE

 Advantages

 Same as internal and external sandbox plus flexibility of

approach in analysis.

9 Prof. Karishma Parab ,Computer Engineering dept,GE

 Disadvantages

 To maintain both an internal and external sandbox

environment.

 It will also be necessary to establish some guidelines on when

each sandbox option is used.

10 Prof. Karishma Parab ,Computer Engineering dept,GE

 Modern Data Analytic Tools
 Basically used to identify patterns and establish relationship
between the data.

 Basic Statistical Concept

 Probability

 Does Sampling and calculates estimates based on sample

 Elements of Inference

11 Prof. Karishma Parab ,Computer Engineering dept,GE

  Statistical model structure

 Data analysis involve how variables are related.

 Bayesian approach

 Supervised classification method

 SVM

12 Prof. Karishma Parab ,Computer Engineering dept,GE

 Reporting Vs. Analysis
 Ultimate goal for reporting and analysis is to increase sales and
reduce costs

 Purpose

 Task

 Output

 Delivery Value

13 Prof. Karishma Parab ,Computer Engineering dept,GE

 Purpose
 Reporting: The process of organizing data into informational summaries in
order to monitor how different areas of a business are performing.

 Analysis: The process of exploring data and reports in order to extract

meaningful insights, which can be used to better understand and improve
business performance.

 Reporting translates raw data into information. Analysis transforms

data and information into insights.

 Reporting shows you what is happening while analysis focuses on

explaining why it is happening and what you can do about it.

14 Prof. Karishma Parab ,Computer Engineering dept,GE

 Tasks

 Reporting Task - building, configuring, consolidating, organizing,

formatting, and summarizing

 Analysis Task - questioning, examining, interpreting, comparing,

and confirming

15 Prof. Karishma Parab ,Computer Engineering dept,GE

 Outputs
 Reporting follows a push approach,

 Reports are given to the users which than extract meaningful

information and take appropriate actions for themselves .

 Three main types of reporting: canned reports, dashboards,

and alerts.

 Canned Reports
 Can be accessed with in the tools.
 They are static with fixed metrics and dimensions

16 Prof. Karishma Parab ,Computer Engineering dept,GE

  Dashboards

 These custom-made reports combine different KPIs and reports to

provide a comprehensive, high-level view of business performance
for specific audiences.

 Dashboards may include data from various data sources and are
also usually fairly static.

17 Prof. Karishma Parab ,Computer Engineering dept,GE

  Alerts

 These conditional reports are triggered when data falls outside of

expected ranges or some other pre-defined criteria is met.

 Once people are notified of what happened, they can take

appropriate action as necessary.

18 Prof. Karishma Parab ,Computer Engineering dept,GE

  Analysis follows a pull approach, where particular data is
pulled by an analyst in order to answer specific business
questions

 There are two main types: ad hoc responses and analysis

presentations.

19 Prof. Karishma Parab ,Computer Engineering dept,GE

  Ad-hoc responses

 These receives urgent request to answer variety of questions ,these

urgent request are time sensitive and demand quick response.

 Short and concise report

20 Prof. Karishma Parab ,Computer Engineering dept,GE

  Analysis presentation
 Some business questions are more complex in nature and require
more time to perform an analysis.

 This analysis results in two key sections: key

findings and recommendations.

 Key finding-This finds out most meaningful and actionable

information.

 Key recommendations – provides guidance on what action to take

based on analysis.

21 Prof. Karishma Parab ,Computer Engineering dept,GE

 Delivery
 Report
 People can access reports through an analytics tool, Excel
spreadsheet, widget, or have them scheduled for delivery into their
mailbox, mobile device, FTP site, etc.

 Automatically delivery of reports on regular basis by computer

system.

22 Prof. Karishma Parab ,Computer Engineering dept,GE

  Analysis

 Humans use their superior reasoning and analytic skills to extract

information from data and provides recommendation for
organization.

23 Prof. Karishma Parab ,Computer Engineering dept,GE

 value
 It’s important to understand the relationship between reporting
and analysis in driving value.
 Value for successful web analytics
 The “Path to Value” diagram

24 Prof. Karishma Parab ,Computer Engineering dept,GE

25 Prof. Karishma Parab ,Computer Engineering dept,GE