janhavi9gupta@gmail.

com
166FVD0TPV

DATA DIGEST

BITS & BYTES

DATA DIGEST
FEB 2021

This file is meant for personal use by [email protected] only.

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

The index
Leadership Speaks 03

Great Learning Success Story 04

The Social Media Bulletin 05

Learner Engagement Initiative 06

Spotlight-Paper
[email protected] publication 07
166FVD0TPV

Alumni Corner 08

Great Learning Journey 09

Employee Corner 10

Let’s Learn 11

This file is meant for personal use by [email protected] only. Page 02

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Leadership speaks
keeps one looking to do better. This is why
overcoming challenges as a team has been
rewarding for me!

2. How do you lead through change?

Change is inevitable and everyone realises it
sooner or later. Staying positive during the
change is a challenge in itself, but it makes the
journey more meaningful. In order to lead
people successfully through change, the first
most essential task is to make people under-
stand the reason for the change as well as make
them feel involved in the transition process. It
SAPNA KAPOOR, serves well to hear out any concerns and
General Manager, Training address these, especially before big changes in
the workplace are to be implemented. The
[email protected] second idea would be to allow for 'settling
166FVD0TPV
1. How do you maintain the company culture time'. While I have always expected only the
and create positivity within the workforce? best results from my team, it is important to
Culture is maintained by people who firmly recognize the aspect of human error, and make
believe in the core values of the company. I space for these so people can learn faster.
ensure that in my frequent interactions with
anyone in the company, I talk about the impor
tance of being a team player, building value in 3. What is that one leadership lesson that you
every thing one is doing, and bringing some- have learnt in your career ?
thing new to the table. There are traditions that Giving and receiving constant and effective
I started when the company's headcount was as feedback has been my most essential lesson as
low as 7 and I am still maintaining these when a leader. Given my role, the ability to share
we are close to 1000. These small (yet impactful) crisp feedback and at the right time ensures
traditions have enabled different individuals to smooth execution of even tricky tasks and
adapt and add to the culture with ease. strengthens channels of communication
between all those involved. In addition to my
Positivity is a very personal thing and can never own journey as a professional, I have seen many
be imposed on anyone. The way I keep my in my team growing into much more effective
workforce positive is by being enthusiastic managers and leads as a result of this.
about everything we do. Encouraging people to
work with passion incites ownership in them
towards their task and brings positivity in their
approach. Also, I have seen, focussing on the
opportunities in all the challenges we face

This file is meant for personal use by [email protected] only. Page 03

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Great Learning Success Story

What other techniques did you use at your
workplace which you have learnt in the DSBA
course?
I would like to talk about the proof of workings
of neurons with the help of binomial distribu-
tion. Like cameras, we have an inbuilt image
stabilising system that makes sure we get a
stable view of the world though we are moving
all the time. This image stabilising system
depends on moving the eyes in the opposite
direction of our head movement. For this to
happen, there are two mechanisms in play. One
depends on the image slippage on the retina of
DR. SRINIVAS DORASALA, the eye, another depends on calculating head
PGP DSBA September’20 Batch
movement and orientation by small gyro
sensors and accelerometers situated in the
Dr. Srinivas is the co-founder of Cyclops inner ears. Anatomical studies have studied the
Medtech. He shared with us how he applied the pathways of both these mechanisms up to the
techniques learnt in the course in his upcoming
[email protected] level of how they move the eyes. These have
166FVD0TPV
product release. This is an amazing testament of shown that the two mechanisms share a final
how the learnings from the program are common pathway.
enabling the learners to get better. And in this During evaluation of multiple patients with
specific case, getting better means helping inner ear deficits, we were observing a correla-
patients lead a better life. tion. The eye movement evoked by optokinetic
stimuli was always more towards the direction
Please tell us what problem you were facing in which the vestibular deficit would move the
at your workplace and how did you resolve it? eyes. This would mean that the neurons are
My company is into eye tracking for diagnosis of 'adding' the two signals (from optokinetic and
balance disorder and neurological disorder. The vestibular). But, I thought what method to
data was traditionally presented as a time series apply to this numerical data in which I am
graph of the eye position against time. Different seeing only two outcomes?
patterns of eye movement related abnormalities This is when the binomial distribution gave the
used to be picked up by studying the graph. I solution. Every event where the direction of
found this to be a tedious task, and also that vestibular nystagmus and optokinetic gain
some patterns would always get missed out. So, being high in the same direction were labelled
I tried to find a solution for this problem and as true. The other situation was false. We found
decided to introduce a new way to represent in a small dataset of 23, that this was statistical-
the information, which has been possible only ly significant. We will be publishing this too with
because of the learnings I have got from the a larger dataset.
PGPDSBA course. I used the kde plot of time I am sure that this is just the beginning, and for
series data against its first differential, to resolve this, I would like to thank everyone at Great
this problem. Learning.

This file is meant for personal use by [email protected] only. Page 04

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

The Social Media Bulletin

During the program journey, the program manager of the batch gets in touch with the learners over
a group video call for an informal interaction and networking purpose. This helps the learners to
know their peers in a better way. Below is the glimpse of what some of the learners posted on their
social media.

[email protected]
166FVD0TPV

This file is meant for personal use by [email protected] only. Page 05

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Learner Engagement Initiative:

Data Science at Work had been envisioned. The client decided to
alter their strategy based on this analysis and
The best way to learn anything is to apply it in launched their apparel store online. The Head
your daily tasks and operations. In your current of Operations at the brand wrote an apprecia-
job, you would have encountered many busi- tion mail to Me for my analytical approach.I am
ness problems - how to automate a repetitive really thankful to GL, who made me understand
task, how to build more efficient processes, or how to apply Data Science in my current role”.
how to exceed expectations in achieving your
KRAs. Once you start looking for such problems Mr Likhit works as a Digital Marketing Executive
around you, Data Science helps you find inno- with an Edtech StartUp. His role constantly
vative solutions that drive significant business demands him to be on top of getting the leads
impact. Eventually, the number of problems that and revenue. One of his key activities is to send
you will solve will help you get the reputation of emails as per marketing campaigns for lead
a data science expert within your organization conversion. The team wasn’t sure on how to
and recruiters’ attention outside the organiza- segment the audience for Email marketing
tion. campaigns and the performance was always
below par.
Data Science at Work is one such initiative
[email protected]
166FVD0TPV
where we encourage our learners to apply He decided it as a great opportunity to apply
data science at work and share their stories what he has learned. He narrated, “ Our data-
with us. Our focus is to help our learners to base has various attributes attached to each
start the journey itself by applying their learning user, so initially, it used to be hard to segregate
in the current jobs and their stories are a great our lists based on user-level attributes. Using K
inspiration for their peers to follow in their means algorithm, I have tried to basket the
footsteps. audience based on user attributes into four
different pools and planned for Email Commu-
Here are 2 DS@work stories from our nication for each basket. And the results have
learners: improved! Open Rates increased from 15% to
Mr Suraj narrated, “one of the clients, a wom- 22%, Click Rates has gone up from 3% to 5%. It
en’s jeans apparel brand, was looking to launch was indeed a happy moment to apply the
themselves on Myntra. They wanted to under- learnings and get the results at work out of it!”
stand what apparel sizes, color, cloth material
etc. is most popular on the platform. I scraped As these data science at work stories are a
the data from Myntra and analyzed it using great inspiration for our learners as well as for
techniques like ANOVA learnt in the program us, we ensure that the best stories get pub-
and suggested a marketing strategy to the lished on our blogs and get shared with our
management. learners on other social media platforms.

The insights were counterintuitive to what the

management had planned. Preferences of
online customers were fairly different to what

This file is meant for personal use by [email protected] only. Page 06

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Spotlight-Paper publication
Electronic Word of Mouth - Decoding feedback given by customers. Mobile holder
Customer Reviews category was used for the purpose of this study
This is a project presented by Payal Upadhyay because it was a utilitarian category and the
and Gaurav Vashisht, PGPBABI students, in the customer opinion extracted can be applied
AICTE Sponsored Online International Confer- across this subcategory evenly.
ence on Data science, Machine learning and its
applications (ICDML-2020). A follow-up paper For the first objective, a list of positive and
was published in the conference journal. negative words were extracted from the reviews
to identify the product features that the custom-
Multiple studies exist where large volumes of ers were interested in. The effect of these words
customer reviews have been analyzed to extract on the product rating given by customers were
useful information regarding their structures, further quantified. Based upon the analysis,
emotion content, readability, syntax, feature these words were further categorized under the
content and ratings to understand the voice of 5-Ps of marketing which could help a seller to
the customer. However, no study exists to better position their product.
establish a link between what the customer
perceives as a helpful feature of a product and For the second objective, it was observed that
how a marketer can exploit the same to make
[email protected] only 5% of the total reviews were displayed to
inroads into the customer community. The
166FVD0TPV the customers on e-commerce platforms like
objective of this study was two-fold, where the Amazon, which were rated and upvoted as
first objective was to help sellers position a helpful by customers. The customer reviews
product for better reach and the second objec- were broken into 50 features and each feature
tive was to identify helpful reviews for custom- was given a numeric score for ease of predic-
ers which will help to make better purchase tion. The voted reviews were penalized to
decisions by analysing the customer reviews of remove biases and to ensure equal consider-
Mobile Holder products on Amazon. For an ation to unvoted reviews. The reviews were
e-marketplace, customer reviews play a very further classified into three slots where 1 was
important role in disbursing information, build- considered as least helpful whereas 3 was
ing trust and promoting commerce. The review considered as most helpful. Through this
data is available in abundance and there is a method, 32% of the reviews were identified as
need to slot, categorize and to extract inference helpful for the customer to make their purchase
from these reviews that can be used to further decisions, which is a significant improvement
the goal of the sellers as well as the buyers. over the industry standard of 5%. The study
further proposed Amazon to add the “Must
The data used in the study was a customer Read” feature apart from the “Top Reviews”
review dataset from Amazon for the sub-cate- and “Most Recent” already available on the
gory of mobile holders, which spanned from Amazon platform. This study can further be
year 2016 to 2018 and the study sample had extended for different types of products. Com-
23,124 reviews with information on 12 variables bining the reviews from multiple social media
which also includes product rating (out of 5), platforms could lead to more helpful insights for
Mayank Sharma VirendraSingh Panwar Kiran Unni
review helpfulness (Yes or No) and qualitative the buyers’ and the sellers’ community.

This file is meant for personal use by [email protected] only. Page 07

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Alumni Corner Things I learnt in Data Science

The journey of Data Science was long. From
revising the basics of statistics to the algorithms
of machine learning, all were taught and
respective assignments were given so that we
can use those learnings into assignments. There
were multiple industry sessions where a known
experienced personality from the industry
discussed things about this field which meant
we got a unique and valuable real-world
perspective on data science. My whole learning
journey was full of discussions, quizzes, group &
individual assignments and the capstone proj-
MANORAMA YADAV ect, which meant that not only was I getting
Data Scientist in Tata Power DDL comfortable with the core subject but also
understanding how to apply them practically.
The final capstone project helped us to imple-
I transitioned my career from a non-data scien-
ment all our learning to solve a challenging
tist role to a Data Scientist role. This article is
real-life problem. All this meant that once the
about my overall experience and my learnings. I
course completed, I was industry-ready.
graduated with a B.tech degree in Electronics &
Communication.. I got placed in an MNC in
[email protected]
166FVD0TPV Challenges during Transition
2014. My first project was in Advanced SQL for
Data Science is an evolving field. New tech-
database management. I worked in the health-
niques are being developed on a daily basis.
care vertical for 5 years, and I discovered I was
Staying updated with the latest tools was an
not learning anything new or exciting on the
ongoing challenge. I did this program along
job. The technology used in the project was
with a regular job. Balancing my time between
very old and there was no scope of using that in
job and studies was challenging. Along with
other companies. Henceforth, I decided that my
studies, I used to devote time to articles and
career needed a new trajectory.
blogs. as well to ensure that I did not lose my
edge.
My career path was not in one direction at that
time, as I learnt Java, I worked on SQL which
Final Notes
was monotonous. In 2017, there was a lot of
If you are ready to make a transition from
buzz around Data Science, Machine Learning &
non-data science to data science field, do
Artificial Intelligence. I started exploring these
proper research, find out your interest level
and the career options associated with them. I
towards data science, and find out an appropri-
searched for the course content, the institutions
ate course curriculum. Devote some time and
and especially the course fees. After looking at
join a structured learning program. Along with
the prospectus of many institutions, I found
theoretical knowledge, get your hands dirty
Great Learning will be the best for me as the
with the real-life dataset and try to solve practi-
course syllabus was modern and comprehen-
cal problems on Kaggle and other platforms
sive. GL had an excellent faculty base and GL is
and share your code on Github.
known for its brand reputation in the industry.
Finally, I enrolled myself for the program. Please read the full article over here.

This file is meant for personal use by [email protected] only. Page 08

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Great Learning Journey

The effectiveness of the training is also mea-
sured using paired sample t-test based on the
scores the students have secured in the apti-
tude test, group discussion and interview,
before and after the placement training.

I personally use sentiment analysis to under-

stand the student expectations and satisfaction
levels from the courses I deliver. This has
helped me to design a better course experi-
ence for our students. Also, I use several
analytics and visualization techniques learned
in the PGP-DSBA program to do result analy-
RENUKADEVI P, sis and report preparation. Apart from my
PGP DSBA March 2020 Batch employment related application, I am using the
supervised and unsupervised models in my
journal publications and time series models for
Renukadevi is from the PGP DSBA March 2020 my Doctoral research program project. The
batch, She is working as a faculty for statistics to
[email protected] course has definitely improved my data
166FVD0TPV
management students in a B-School, and is handling and analytics skill, that has given me
pursuing PhD in Finance. She says that her confidence in my course delivery for my
journey with GL is incredible and here is a brief students and has improved the quality of my
of our conversation with her. research work.

Renukadevi says, “we provide placement I take this opportunity to thank GL and the
support to our students. We observed that the mentors who guided us throughout the course”
expectations of the recruiters are unique and
specific with respect to the roles. But our
placement training was a generalized one,
where we train students to take up aptitude
tests, participate in GD and attend personal
interviews. So, we planned to design a place-
ment training for the final year students. To
understand the requirement of the recruiters,
we decided to get the inputs from their web-
sites and the posts in social media. As this data
was huge, we did text analysis to find what
were the most required skill sets by the recruit-
ers. Based on that, we designed the custom-
ized placement training based on the student’s
background and interest. This helped us to
improve our placement percentage by 24%.

This file is meant for personal use by [email protected] only. Page 09

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Employee Corner
This time we have Ritesh Sethi and Saibi Gambhir. We asked them the following 2 questions and
this is what they had to say

Ritesh Sethi Saibi Gambhir

Manager, Operations Assistant Manager, HR

What do you like most about your work? What do you like most about your work?
There cannot be a single answer to this. Since I Acquiring the best talent in the market and
have joined the company, I have enjoyed the seeing them grow within the organisation, the
ride to the fullest. If I have to pick up only one feeling of building up a team by hiring a good
aspect, it is the opportunity to contribute to the
[email protected] number of resources and then seeing them
cross functional teams. Being the SPOC of the
166FVD0TPV being great performers is an incomparable
guru community has given me that opportunity emotion, full of satisfaction and pride. Apart
to engage Gurus to numerous verticals, like from this, as part of employee engagement,
marketing, branding, networking etc. and not maintaining the basic hygiene in the organisa-
just restricting the engagement to operations. tion through #FunatWork activities are some-
The sheer sense of satisfaction to contribute to thing I enjoy the most.
the multiple growth aspects, and actually create
impact, is something that stands out for me. What was your most recent memorable
experience at work?
What was your most recent memorable When you work in human resources manage-
experience at work? ment, you set an example and you have much
I have 2 such instances here. The top most was influence over how people are treated and the
the appreciation that the DSBA guru community company culture. You can tell your colleagues,
got from the senior leadership team for the coworkers, and employees how much you value
impact the gurus have had in cross functional them and their contribution. The virtual RnR
initiatives. It feels amazing to see such acknowl- felicitation ceremony that happened for the
edgment. Second one had to be the selection sales team became a memorable one. That is
of 14 papers from the BABI track for the ICD- because of the happiness and excitement
ML20 Conference. It was one of the most showcased by both the employees and their
critical initiatives I was a part of and felt managers.
outstanding to see our learners' efforts being
recognized at the biggest stages.

This file is meant for personal use by [email protected] only. Page 10

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Let’s Learn

Using Predictive Analysis, companies can

identify such claims faster.
Author:
Swati Deval Data Analytics for Customer or Insurer
Most companies provide self service portals to
customers. It helps them to manage various
Data Analytics in Insurance policies and submit claims. There are indepen-
dent websites which compare insurance prod-
Every person wants to protect his/her family and ucts from different companies and suggest best
self against unexpected risk. Insurance industry fit products based on customer need. This helps
caters to this need by guaranteeing payment for customers to choose policy with minimum
uncertain future events in exchange for a premi- premium or get some additional cover.
um. Depending on which uncertain event insur-
ance companies cater to, it is classified as life, Enabling the Future with Data Analytics in the
health, property, etc. Managing risk is a major Insurance Sector
purpose of insurance companies and hence
[email protected] Machine Learning and AI algorithms along with
historically they were using mathematical and
166FVD0TPV
vast data collected by Insurance companies are
statistical models to analyse data to decide changing the way of working for the Insurance
premium for the products. With the progress of sector. Many health insurance companies
the data storage facilities, the Insurance sector encourage customers for healthy habits. Compa-
also started collecting data in large amounts, and nies are able to identify low risk and high risk
data analytics became a day to day affair for customers. In turn they can reward low risk
them. There are many stakeholders in the insur- customers by reducing premium, and can
ance sector, and each of these stakeholders improve risk analysis for underwriting high risk
benefits from the use of Data Analytics. customers.

Data Analytics for Insurance Company Auto Insurance companies also give smart
Underwriting, claims and finance are major devices to install in cars for which policy is
departments of an insurance company that uses purchased. This device collects information
data analytics to improve company performance. about driving habits, maintenance of cars which
They collect information about their customers, can be used in risk assessment for next year
and identify the various segments. Predictions policy and rewards good drivers.
can be then made based on these segments
about behaviour of customers. Underwriters So not only past data, but real time data
assess risk in granting policy to a customer and collected during policy tenure will be used by
decide premium based on identified risk. Fraudu- insurance companies to improve their sales and
lent claims are a major challenge for the claim profit. They will also provide a better customer
process. To find out if a claim is genuine or experience, affordable premium to their
fraudulent manually is very time consuming. customers.

This file is meant for personal use by [email protected] only. Page 11

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

Discover native models and Generative models. To under-

stand GANs, we should know about generative
Understanding consulting roles models and how they are different from Discrimi-
The demand for Consulting roles is always high, native models. Discriminative models classify
even in a tough economic condition. The kind of input data; i.e., given the features of an instance
exposure, travel, and knowledge it provides is of data, they predict a label or category to which
admired by many individuals across the globe. that data belongs. In Supervised Learning, the
While the demand is high, naturally the aspira- classification algorithms/models are examples of
tions will also be very high. So, it is highly recom- discriminative models.
mended for any aspiring consultant to spend a Generative Modelling is an unsupervised learn-
good time in preparing oneself. This article ing task in machine learning that involves gener-
mainly focuses on developing many skills that are ating new data samples from the probability
not like the usual process of getting a degree, distribution of training data. Given some data,
earn a work experience, and look for openings the aim is to have a model for the underlying
with leading consulting firms. This article focuses probability distribution of that data so that we
on an effective process that will put you ahead of can draw samples that are similar to our training
the queue. data.

Data Science post covid 19 Learning rate in Machine Learning

Life during and after COVID might be relatively In supervised learning, to enable an algorithm’s
similar. Technology, and in particular the internet predictions to be as close to the actual values/la-
[email protected]
& telecom/ISP is perhaps the only source of
166FVD0TPV bels as possible, we employ two things:
communication, socializing, working and almost 1) A cost function and
everything else. There are two sides to COVID’s 2) A technique to minimize the cost function.
impact. One to many, and many to one. In this
backdrop, Data Science is an integral part of There are popular forms of cost functions used
businesses, daily routines in personal lives, and for different tasks that the algorithms are expect-
maybe a new way of networking. So, let us focus ed to perform. Also, a popular technique used
on some immediate components and look at to minimize the cost function is the gradient
each sector independently and look at possible descent method. We will understand these
scenarios “During Lockdown”. concepts to understand the role of ‘learning
rate’ in machine learning.
Transition in the same company
If you want to be in analytics then the First part is
an opportunity and if you are working in such a
company then your internal transition into analyt-
ics is an easy gateway. This article is an attempt
to streamline important areas to focus, key skills
to develop and right doors to knock. Hoping
that you have built your reputation already.

Understanding GAN
The learning models in machine learning can be
classified into two sub-categories, viz – Discrimi-

This file is meant for personal use by [email protected] only. Page 12

Sharing or publishing the contents in part or full is liable for legal action.
 DATA DIGEST
FEB 2021

[email protected]
166FVD0TPV

Editorial Team
Program Managers - DSBA Operations team, Great Learning

Prabhjot Kaur Saakshi Nayak Richa A. Sethi Faisal Noor Khan

This file is meant for personal use by [email protected] only.

Sharing or publishing the contents in part or full is liable for legal action.