END-TERM EXAMINATION
Program Name PGP Academic Year/ Term 2022/IV
Course Code EIA121401 Course Name Data Analysis using Python
Total Marks 40 Time 2:00 Hrs
Regn. No.: 2021PGP220 Santoshi Devi Nayudubathula
Instructions:
Open Yes Open Laptop Ye Internet Ye Calculato Ye
Book s Access s r s
Other Information:
1. A Participant has to answer all questions.
2. A Participant has to submit Jupyter notebooks/Python code files with the question number
as a title along with registered number wherever needed. There is no need to write the
same code in answer sheets.
Q1. Suppose you are working as a Marketing Manager in an MNC. You found that [Marks 5]
the marketing campaigns/advertisements are not improving product sales.
How, you can use Data Analytics to improve product sales.
Ans 1.
1) Improve value propositions and price points:
I will make sure that I am constantly saying the correct thing, to the
right customer, at the right time by knowing how to analyse sales
data.
Building highly-individualized value propositions suited to the
particular requirements of each customer segment is feasible
through the collection and cross-referencing of sales data analysis.
Pricing new goods and services at a level that will generate the most
revenue is another difficult task.
I will test a variety of price points to establish what the best price is
for each solution, and even for each customer segment, by
employing market data, dynamic pricing engines, and sales data
analysis skills.
2) Narrow and Refine Product offerings:
Transaction analysis allows me to identify products whose sales are
underperforming overall or in certain consumer segments.
I will reduce costs by discontinuing these underperforming goods
and concentrate more time and resources on those that generate
1
the highest sales and profits.
3) Disruption and Innovation:
I will swiftly identify client needs with the help of sales data
analysis, and I will provide individualised solutions to customers
more affordably than my rivals.
4) Accurate Sales Forecasting:
Contrary to setting lofty goals, historical data provides me with a
precise, realistic view of how much my team should make over the
course of a specific time period.
5) Performance Analysis and Incentive Plans:
Managers may improve performance concerns, create attainable
sales targets, recognise and reward top performers, and inspire
their team by using sales data analysis.
6) Increase Repeat purchases from Existing Customers:
Reaching out to clients to cross-sell and up-sell products I believe
could also fulfil their needs is the first step in boosting revenue
quickly.
A quick and long-lasting strategy to increase revenue is to use data
to locate underserved clients and then make tailored cross-sell or
up-sell recommendations to them.
7) Targeted marketing to cut costs and Increase ROI:
I will efficiently "clone" my most profitable clients and save money
by using data to increase the targeting of my advertising efforts
rather of spending it on people who aren't likely to be a good fit for
your business.
Q2. HBR Guide suggests six key steps for analytics-based decision making. Explain [Marks 5]
these steps with an application from your interested domain.
Ans 2.
Six key steps for analytics-based decision making from HBR is mentioned
below. The domain I would choose is Product Management:
1) Recognize the problem or question:
Frame the decision or business problem, and identify possible
alternatives to the framing.
In the product management domain, the first step would be to
recognize the problem the users are facing, finding the gaps
between the current offerings and the expectation from the
consumers.
2
Asking right questions from the consumers, what they expect from
the product, what is the problem they want to solve using the
product and the impact of this gap in the current situation.
This would help us to frame the right problem and move in the right
direction. We would not deviate from the actual problem at hand if
the problem statement is understood properly.
For example:
Before OLA was launched in India, the founder had faced the gap in
the current transportation system, he could sense the problem the
consumers were facing and build the right product based on the
Indian market. This led to success of the company though the rival
UBER had a presence in the market.
2) Review previous findings:
Identify people who have tried to solve this problem or similar ones—
and the approaches they used.
This step is vital to know the product-market fit and the approach used
by the competitors to launch the product in the market.
The data would help us know the area where the competitors lagged
behind and the reason of their failures or success.
For example:
Spotify did not enter the Indian market as late as 2017 the reason being the
company was analyzing the existing player’s strategy and the necessary
infrastructure available in the Indian market. The previous findings helped the
company analyze the competitor’s success and failure reasons and helped
company strategize the best way to enter the Indian market.
3) Model the solution and select the variables:
Formulate a detailed hypothesis about how particular variables affect the
outcome.
After analyzing the past data and gathering all the customer expectation
data, we would get idea of the direction we should take to solve the
problem.
We would know the variables that would be crucial for the success of the
product. In the product management domain, customer feedback,
competitor’s analysis, market conditions, gaps in the market, user pain
points, etc. are gathered from the market research, post which the
hypothesis is made how the above factors would be helpful for the success
3
of the product.
For example:
OYO before the launch had done a thorough market research and reached
out too many hotel chains for the business model. They made a hypothesis
that the consumers would book the hotel stay online and would trust the
whole business
5) Collect the data:
Gather primary and secondary data on the hypothesized variables.
This is the most vital part of the process. Data collection would begin right
from the idea generation to the launch of the product, and post launch of
the product as well.
User feedback, user interview about the product, getting feedback of all
the stakeholders involved in the process, would give a wide idea about the
product feedback.
6) Analyze the data:
Run a statistical model, assess its appropriateness for the data, and repeat
the process until a good fi t is found.
Once the data is collected it should be analyzed to get a fair idea about the
business and the consumer behavior.
The buying patterns can be analyzed and the company sales can be
impacted by this data
7) Present and act on the results:
Use the data to tell a story to decision makers and stakeholders so that
they will take action.
Once the data is collected and it is analyzed, further insights can be found
out from the data.
A story can be prepared from the data to get a high level understanding of
the data. This would help the decision makers to make better decision that
would be beneficial for customers, employees and business.
4
Q3. Exception handling is an important concept in Python to handle run-time [Marks 5]
errors or exceptional conditions. Explain the exception handling mechanism of
‘FileNotFoundError’ using Python.
Ans 3.
We will use try-except block to handle FileNotFound exception. The try-block
will have the code to be monitored for any exception. On the other hand the
except-block will contain the code to handle the exception occured.
The Source Code – Python Read and Show File Program with Exception
Handling
Python Exception Handling FileNotFoundError
If we try to open a file for reading which does not exist
Then a FileNotFoundError exception occurs and we will display our user
defined custom error message. We have also imported the sys module. This is
beacause we will use sys.exit() function to terminate the Python program if
exception occurs after displaying a suitable messae to the user.
The output:
Sorry file not found
Why To Handle FileNotFound Exception?
This program may be written without handling the exception, but in this case,
the program will terminate and will show default error message. The default
messages may be hard to read and understand.
5
Therefore, we use the exception handling in this program.
Q4. Exploratory Data Analysis (EDA) helps in exploring the data using statistics and [Marks 5]
data visualization. Give three possible times when data visualization does not
work?
Ans 4.
1) India Today, Chances of NDA Coming into Power
Ahead of the 2019 elections in India, India Today published an article to discuss
the chances of Prime Minister Narendra – NDA – Modi winning a second term.
While NDA did win his second term, understanding his chances via this visual
data is puzzling, unless all India Today readers are professional data analysts.
It’s ideal to use a speedometer chart in data visualization like this one (or a pie
chart in general) when you have collectively exhaustive and mutually exclusive
quantities. In this chart by India Today, none of the two is the case.
The chart is divided into 3 – NDA staying below the 220 mark, NDA crossing the
250 mark, and NDA getting a majority. The probability is 9%, 72%, and 50%
respectively.
Commonly, a speedometer chart has just one pointer. However, the graphic
designer here decided to feature two pointers – one points in-between the 9%
and 72% probability, while the other points at the 50% probability.
An ideal explanation here would be that the chances of NDA winning a second
term fall on the 72% probability as it’s in between the pointers. However,
there’s no way to be sure.
2) Dialect Map of India
English and Hindi are the two state languages in India. However, there are
about 121 recognized languages out of 1,369 recognized dialects. This dialect
map of India aims to represent this data visually via a Choropleth.
A Choropleth is a good way to visually interpret data, but it becomes unreliable
when there is so much data to interpret. That’s what this map designated
6
failed to perceive. The map features 60 Indian dialects and others. Hence,
there are so many colors, which makes the data representation appear as a
color riot.
You can’t tell which dialect is the most dominant by looking at this map.
Matching the colors to the top languages will also be challenging as colors of
close shades are featured next to each other.
The black background makes things a bit more complicated as it mixes with the
black colors featured on the map.
Choropleths are best for representing how variables change across different
areas. When it comes to population distribution, which this map is all about,
choropleths are not ideal because it results in uneven distribution.
The Economist, Why ticket prices on long-haul flights have plummeted
The Economist isn’t a publication one would expect a bad data visualization
example from but here’s one. The core here is the featured protractor.
One would expect to grasp a relationship between long-haul flights and their
ticket prices. However, what you get are overlapping lines such that
differentiation is a problem.
The protractor features two lines. The blue-colored lines are for transatlantic
flights while the ash-colored lines are for other flights. There are three axes
with two being Distance in Km and the other, Change in price of economy-class
tickets.
Since the length of the lines depends on the flight’s distance, some lines are
terse. Hence, it’s difficult to trace them to grasp the percentage change in
ticket prices. Furthermore, the actual prices of the flight tickets are not pointed
out.
Except for their not-so-straightforward titles, the three graphs featured below
the page are easier to understand. The first represents shares of Norwegian
seats on six transatlantic routes, and the second represents jet fuel $ per liter.
In comparison, the third represents average ticket prices on six transatlantic
routes.
Q5. The Human Development Index (HDI) is a statistic composite index of life [Marks
expectancy, education, and per capita income indicators, which is used to rank 10]
7
countries into four tiers of human development. Using Python, clean the data,
analyze global HDI data (CSV file is given) and report interesting insights. You
can assume possible 3-5 questions on the data.
Ans 5.
Q6. Collect recent tweets about Climate change in India. Analyze the sentiments [Marks
from these tweets using NaïveBayes Analyzer of Python. Plot Positive and 10]
Negative sentiments using WordClouds.
Ans 6.
