Case Studies

In general, there are two types of questions which are asked in an interview:-
1. Business Case Study
2. Guesstimate

First of all, let me do the mandatory promotion of the e-books
1. Case Studies and Guesstimates for Data Science (Recommended by the Readers)
2. 100 Puzzles and Case Studies to Crack Data Science interview

Let’s solve some Case Studies and Guesstimates. This will also give you a glimpse of the type of questions which are asked in interviews 🙂

1. How much is the annual income of a beggar in Bangalore(Mynta – Second round interview)

You can assume anything and everything under the Sun, just to try to keep the assumptions close to reality
I always start with an equation, for this question the equation which I assumed was:-

Amount per day * Number of Calendar Days (365)

Assumption 1- A beggar begs all day of the year
Now, I have divided a complete day in 4 parts
6 am to 10 am – High income
10 am to 4 pm – Low income
4 pm to 10 pm – High income
10 pm to 6 am  – No income 

Assumption 2 – The beggar will bet more money in slot 1 and 3
Assumption 3 – Beggar interacts with 500 people in each slot
Assumption 4 – The success ratio table

Slot	Success Rate	Number of people giving money
6 AM – 10 AM	0.03	45
10 AM – 4 PM	0.01	15
4 PM – 10 PM	0.05	75
10 PM – 6 AM	0.006	9
		144

Assumption 5 – Probability of amount, I have safely assumed that 30% people will give Rs.2, 20% will give Rs.5 and 50% will give Rs.1

Slot	Success Rate	Number of people giving money	Amount
6 AM – 10 AM	0.03	45	94.5
10 AM – 4 PM	0.01	15	31.5
4 PM – 10 PM	0.05	75	157.5
10 PM – 6 AM	0.006	9	18.9
		144	302.4

Now we have Rs.302.4 per day income.
Annual amount = 302.4*365 =  Rs. 110,376 It doesn’t matter if the amount is high or low, what matters is that you have an approach to solve the problem. Few more things which you can add here are:-
1. Divide the year into seasons
2. Divide year into weekend and weekdays
3. Public Holidays

2. How many cars are there in Bangalore ? (Myntra Case Study Round)

Let’s directly jump to the approach part

Assumption 1 – Population of Bangalore, 10,000,000. If you are talking about a metro city in India then assume something between 5 Million to 15 Million. In case it’s about a town, then assume something around 1-5 Million.

Assumption 2 – Each family has 4 members, so the number of families will be 2.5 Million. From here you have to go concrete with your numbers

Now divide the families into 4 groups and assume a percentage distribution of the population(as a family)

Income category	Percentage of Family
Lower Middle Class	50
Middle Class	30
Upper Middle Class	15
High Class	5

Now we can assume the number of vehicle for each category

Income category	Percentage of Family	Number of vehicles
Lower Middle Class	50	0
Middle Class	30	1
Upper Middle Class	15	2
High Class	5	5

Assumption = Middle class have only two-wheeler, upper middle class have one two-wheeler and one four-wheelers, the high class will have 4 four wheeler and 1 two-wheelers. So we can now calculate the number of two and four-wheelers in the city

Income category	Percentage of Family	Number of vehicle	Two wheeler	Four wheeler
Lower Middle Class	50	0	0	0
Middle Class	30	1	750000	0
Upper Middle Class	15	2	375000	375000
High Class	5	5	125000	500000

You can add other parameters like:-
1. Number of cabs
2. Divide the whole population into 2 group, cab drivers and others. Assume that 10% are cab drivers and 40% of these cabs are only bought for rent purpose. This way you will have =(((2500000*10)/100)*40)/100. Here we have also assumed that there is only one cab driver in a family

3. Company Name – The Moonfrog
Round – 2 (Case Study)

The case study round was more about solving any real-life business problem. I was asked about my hobbies and I answered with “playing Clash of Clans”, the case study topic:– 

“How will you change the UI of the game to increase the number of people buying coins from the shop section”

This was a business case study, we had the discussion on the following points:-
1. Project the offer directly on the home page instead of clicking on the shop button
2. Put the price to buy elixir and gold near the collector(the pink and yellow images)
3. After every attack resulting in a loss, give an option to buy back the points
4. Get the country-wise data and see the most engaging time of the player and project the offers accordingly
5. Give a variable discount to the players on the basis of their day-to-day performance to allure them in buying coins on the day they performed well

4.
Company Name – Myntra 
Round – 2 (Case Study)

Topic:- Suppose you have a restaurant and there is a crunch of sitting space, apart from that you have one kitchen door which is generally crowded with the waiters and food delivery guys which leads to delay in serving the food. Would you like to create room for one more small window to expand the kitchen and make the service quicker?
What will be the impact on the already small sitting space?
How to optimize the above scenario?

Approach:- Throw numbers wherever possible.
If I were the owner of the restaurant, I would have added a small room for some selected food items. Looking at the data, we can conclude on a few food items which are served most often to the dining customers as well as to the delivery customers.

Suppose it’s a Punjabi restaurant and the best sellers are Naan, Butter Chicken, and Punjabi Biryani. Once we have the data, we can have these items prepared in the main kitchen, but the delivery/serving should be done on the small room/window which is newly created. What about the space crunch? The waiters will have a hard time going back and forth?
The space crunch needs to be compromised and in return, we will be providing quicker service which should compensate for space. The serving time and sitting time of the customer will decrease which might counter the space crunch as more number of customers can be entertained. 

There is one more problem with the restaurant and that is the wastage of raw materials as the shelf life of a few ingredients are very low. How to counter it?

Looking at the data we can come up with a few nodes of this problem. We can ask these questions to the data:-
1. Which all items are sold the most?
2. What is the key ingredient of these items?
3. Day-Food item pair, with this we can get which all items are sold on which day

We can predict the amount of demand in the coming week and can act accordingly. But prediction might not work the best in every case, right?
I agree, to counter this we can have a variable price menu, so if the shelf life of a particular item is low, we can give a discount on these items or can give a combo offer to clear the stock at the very end of the day. There were questions on the formula which you will use to determine the new price in the above situation. Basically, you need to come up with parameters and you have to decide the importance of the parameters by either giving them multiplicative or additive importance in the formula or you can club your answer with any other offer. Following were the discussion points:-

1. We can take those items which will get wasted by the end of the day(looking at the stock near the closing time) and can set a variable price. Suppose, you realize at 9 pm that at least 100 kgs of Biryani will be wasted tonight(by 12 am), then you can set a variable price on Biryani
Cost Price of 1 kg of Biryani – 100
Selling Price of 1 kg of Biryani – 200
Stock left – 100 Kgs
Time left – 3 hours
Aim – To minimize the loss Revised Price = ((180 – x)/180)*Selling Price
where x is the number of minutes after 9
So, at 9:30 pm, the price of the Biryani will be = ((180-30)/180)*200 = Rs. 166

This is one way where we can help the restaurant in clearing the stock.

We can also give the customers a Mega Offer, where if they buy a Biryani today, then they will get some y% discount tomorrow. This will also help in retaining customers.
Apart from that, we can also give them more Biryani at a lesser price. Suppose Rs. 166 is the price at 9:30 pm, then we can offer them 1.5 kgs of Biryani at Rs.210. This will clear the stock easily.

5. Company Name – Xiaomi
Round 3 – Case Study
Topic – Profit of a company selling mobile back cover is declining. List out all the possible reasons

Following is the way in which discussion proceeded with the interviewer:-
1. The demand itself has declined i.e. customers are not using cover that much. Asked to think more by the interviewer
2. Maybe the competitor is also facing loss which again means that the demand is low. Competitors are making a decent profit
3. Bad Marketing – The company is not putting stalls or shops in a crowded place. The interviewer told that the company was making a decent profit 6 months back
4. Maybe the footfall of the mall or place decreased. Could be(first positive response)
5. Maybe a popular mobile phone shop has shifted somewhere else. Could be(again a so-so response)
6. Maybe the other companies have reduced the price of their product which is why customers are drifting to these companies. The interviewer seemed pleased
7. New technology in the cover market to make covers more durable and the company we are talking about is using the same old technology. Seemed good enough point
8. Since we are talking about back covers, there could be new or trending designs which are not produced by the company
9. The company has not registered on different e-commerce websites and the website they are present on is not doing good business. He looked satisfied with the point

Find more Case Study and Guesstimates on the books mentioned above and the links given below:
1. Number of Smart Phones sold in India per year
2. How many laptops are sold in Bangalore in one day?
3. How many T-shirts are sold in India per day?
4. Ola case Study

We will keep on adding contents here.

The Data Monk