Welcome to the first module of interview questions. In this module, we would be looking at what are guesstimates, why are guesstimates asked in a data science interview, how to approach guesstimates, few practice exercises of guesstimates and in the end we will look at tips to solve guesstimates. Without wasting much time, let's start with what are guesstimates. In this video, we would be looking at what are guesstimates. As the name suggests, it's an estimate based on a mixture of guesswork and calculation. In guesstimate you need to estimate a quantitative variable that is relevant to the case you are solving. The case might be anything. It might be relevant to the job you are applying for. It might be relevant to your personal life or your city you are staying in. It can be anything. Let's understand it with an example. Here the question is, how many AC are installed in your city? So here, the interviewer might not also know the answer to this question. He might not even know your city or your country. But still he's asking this question. You must be thinking, why? In the next video, we will be looking at why are these questions asked in a data science interview. In this video, we would be looking at why are guesstimates asked in a data science interview. The interviewer will give you the question and will give you only a couple of seconds, then he'll ask you to speak up. He does this due to various reasons. One of the reason is he wants to put you in a stressful situation and wants to understand your ability to deal with stress. He further wants to understand how structured is your approach, or how well are you able to structure your approach. It will give him confidence that if you are able to approach the problem here, you would be able to solve complex data science problems in the real life scenario as well. While thinking about the approach, you would be having different methods in back of your mind. How fast are you able to make quick checks on the efficiency of different methods would also be a check here. The next piece he wants to understand is your hold on mathematics. He wants to see how comfortable are you with the numbers and how fast are you able to do back of mind calculations and validate the magnitude of numbers. In short, he'll get a lot of idea about you and the non-technical background with just one question. Now, when you have understood why are guesstimates asked in a data science interview, now let's look at how to approach a guesstimate. In this video, we would be looking at how to approach guesstimates. The first step is to understand the question or to clarify. Try and extract the exact information from the interviewer on what he wants you to calculate. If we take the previous example, it's not clear which AC is the interviewer talking about. As AC are installed in houses, vehicles, offices, industries, etc. Clarify if the interviewer wants to know total AC in the houses, or it includes vehicle AC, industrial AC as well. Always ask first before you attempt any guesstimate. The next step is to devise a logical approach. There is no foolproof way to approach a guesstimate. You can solve a guesstimate using several approaches. The trick here is to go with the approach that helps you minimize your assumption. It's a simple rule of thumb while you approach guesstimates. In this example, devise multiple approaches, be it starting with population of the city or area of the city. Consider all the possible factors which might be useful in calculations, like weather, population, income of the city, etc. The next step is to decide which approach to take. We call it analyze step as well. Once you have devised the approach, it is time to decide which approach you want to use. Ideally think of two, three steps ahead in the approach you decide to use and see how it works. The final step is conclusion. Put all the numbers and assumptions in place. This is the final step. Take the approximation in such a way that it becomes easy for you in the next step. Once you get this right, you have cracked the guesstimate. However, once you practice enough, you develop a knack of solving it. Welcome to the first question of guesstimates. Here the question is, how many AC are there in your city? I'll take a pause here. See if you can attempt the question. Let's move to the solution now. So here the question was, how many AC are there in your city? So first thing I need to specify is which is my city. So my city is Bangalore. It's the IT hub of India. Now we need to understand that what are potential users of the AC. There are lot of potential users. It can be household, it can be vehicle, it can be industry, it can be office, it can be anything. We need to clarify this from the interviewer, which AC is he talking about? So here, let's assume that he is talking about household AC. So assumption one is that the AC we are dealing with are household AC. Are household AC. The next step is to devise a strategy or an approach. So population looks like a good starting point. The population of Bangalore in 2011 was around 84.3 lakhs. Okay. So, let's take a second assumption, that is in 2020, that is the current year, the population is 1 crore. This is the current population. This is the population. Since Bangalore is the IT hub, we can assume that 80% of the population is tenant, and only 20% are landlords or permanent residents. So let's devise it that 80% population is tenant and 20% population is landlord. Okay. So here 80% of 1 crore is nothing but 80 lakhs. Here 20% of 1 crore is 20 lakhs. Out of these 80 lakh people, few would be staying in double sharing, few would be staying in single sharing. Let's assume 50% are staying in double sharing and 50% are staying in single sharing. These are double sharing and 50% are in single sharing. The total rooms required for these 50% people would be 50% of 80 is 40 and they are staying in double sharing. So it comes to be 20 lakh rooms. And for these 50% of the people who are staying in single sharing, the total number of rooms would be 40 lakhs. So out of the tenants, we have around 60 lakh rooms. Let's look at the number of rooms from landlord. Let us assume that for 20 lakh people, 15 lakh rooms are required. Now, the total rooms in Bangalore are 60 lakhs plus 15 lakhs, that is 75 lakhs. Since the weather of Bangalore is quite good, there isn't a need of AC. So assume that out of 60 lakh rooms of tenants, only 30% of them have AC. This brings down it to be 18 lakhs. That means 18 lakh AC are installed in the rooms of tenants. And out of the 15 lakh rooms where landlords are staying, let's assume 50% of them have AC. This brings it down to be 7.5 lakhs. So now, the total number of AC installed are 18 lakhs plus 7.5 lakhs, that is 25.5 lakhs. 25.5 lakhs plus there might be few AC installed in the hall as well. Let's take that to be 4.5 lakhs, that is total 30 lakh AC are installed in Bangalore city. This is the final answer, but as much the answer is important, this might not even be correct. Total number of AC might be 40 lakhs or might be 20 lakhs. But the interviewer would be looking at the approach which you have taken or the numbers like how is the guess? So here 30% tenants have AC in their room but 50% landlords have AC in their room which is quite justifiable, right? The landlords would be staying there for a quite longer time. So they are in a good position to invest in AC, while tenants won't be preferring to invest in AC. So this is how you should structure and you should approach a guesstimate. Welcome to the next question on guesstimates. The question is, how many red cars are there in India? I'll take a pause here, see if you can attempt the question. Let's look at the solution now. There are a lot of ways to solve this problem. I'll start with population. The total population of India is 130 crores. The population is 130 crores. And an average family size in India is four. So the average family size here I'm taking this assumption that the average family size is four. So the total number of families, total families in India is 130 by 4, that is 32.5 crores. Let's approximate it to 30 crore for the ease of calculations. So in India, we have total 30 crore families, out of which few would be lower class, few would be middle class, few would be upper class. Let us differentiate the purchasing power of families with their economic status. So out of 30 crore families, few families would be lower class, few would be middle class, few would be upper class. Okay. So if you take a guess, 30% of the families are lower class, 60% of the families are middle class, while 10% of the families are upper class. So total number of lower class families would be 30% of 30 is 9 crores. Total number of middle class families would be 60% of 30, that is 18 crore families. And upper class families would be 10% of 30 crore, that is 3 crore families. All the lower class families, let's assume that they won't be able to afford any car. Out of middle class families, few would be able to afford a car, few would not. So let's assume that there are 50% of the families which are not able to afford any car. Okay. Then there are 30% families which are able to afford one car. Okay. Then there are 20% families which have afforded two cars. Okay. These are slightly upper middle class. Now, if we look at the upper class, let's assume that all the families have two car each. Few families would be having three cars, but few would be having one car as well. So I've taken the approximation of two car. If I look at total number of cars in India, then it comes out to be 5.4 crores plus 7.2 crores plus 6 crore. Okay. So the total number of cars in India is 18.6 crores plus there might be 1.4 more crore cars in form of Ola, Uber or some rental services. The total number of cars in India is 20 crores. Okay. So now we have solved half of the problem. That is how we got how many cars are there in India. Now we need to find how many red cars are there in India. So in India, there are total seven major colors. Seven major colors of car. Okay. Out of these seven colors, white and black are the most famous colors. White and black. So let's assume that these two colors comprise of 50% of the market. Okay. That means 50% of the market of 20% is 10 10 crore cars. Now, we have five colors and we have 10 crore cars. So out of it, one color is red. So red color is nothing but 20% of 10. That is 10 by 5. Okay. That is 2 crore cars. So in India we have 2 crore red cars. This might not be the exact answer, but you got an idea about the approach, right? The interviewer would be sold out at this answer. Let's look at the practice problem now. The question is, how many weddings are performed each day in London? So you need to predict or guess the number of weddings that are performed in London each day. Give it a shot. In this video, we would be looking at tips to solve guesstimates. Analyze all possible uses of the subject. For example, while estimating the number of AC installed in your city, you should consider AC being used in houses, commercial buildings, restaurants, vehicles, and all other places which are potential users of the AC. Here AC is the subject and the potential users are houses, commercial buildings, restaurants. The next tip is to keep the population of your country, state and city on fingertips. As population is the most common proxy for many case studies. Such numbers give you a good starting point. This is personally my favorite. I try to approach the problems starting with the population of the city or country itself. The next one is to draw neat diagrams to show the segmentation. This not only helps to calculate quickly, but also makes it easier to redo the calculations on the segment level if required. Ideally, you should not go back and redo all the calculations, but if required, if due to some reason the approach failed, then the neat diagrams will help you to approach the problem in a quick way. The final tip is don't do round offs in the same direction. Such round offs magnifies the error term. Putting a sign in front of the rounded off numbers help. With this, we come to the end of this module. I hope by now, you would be having a better idea about how to solve guesstimates. I'll see you in the next module in the next video.
