Welcome to the SuperDataScience podcast. My name is Kirill Eremenko, data science coach and lifestyle entrepreneur, and each week we bring inspiring people and ideas to help you build your successful career in data science. Thanks for being here today and now let’s make the complex simple.

 Hello ladies and gentlemen and welcome to the SuperDataScience podcast. I just literally got off the phone with Neelabh Pant and this episode blew my mind completely. I really hope it’s going to blow yours too because Neelabh has so many stories to tell about data science. We had so many laughs and as you probably noticed this is a much longer than usual episode just because I didn’t want to stop talking. I didn’t want to stop talking to Neelabh, it was so great. Neelabh has worked in geospatial analytics, he’s combined that with machine learning and deep learning experience, he’s worked with genetic algorithms. And the most interesting thing is he’s got so many crazy ideas of how to apply that in business, he actually applies it business and he applies that in real life. Well, I’m not going to ruin it for you, you have to check out this episode. Whatever level of data science you’re in, this will get you pumped about data science. Let’s get this thing rolling without further ado, I bring to you Neelabh Pant.

[Background music plays]

 Welcome ladies and gentlemen to the SuperDataScience podcast. Today I’ve got a special guest, Neelabh Pant on the show, calling in from India. Neelabh, how are you going?

Neelabh: Pretty good, how are you?

Kirill: I’m going well, thanks man. And right off the bat, how did we meet? Tell us the story; we were just chatting about it just now, it’s such a fun story.

Neelabh: It’s exciting. It’s just an amazing story. As a matter of fact, I really got into data science so much so that I just learning from online resources wherever I could find. For the first time, I saw Kirill was on YouTube for one of his promotional videos for SuperDataScience.com and then I started following him there, started reading his courses like data science, machine learning, Tableau and I was like, man I just want to see this guy because this guy is amazing. I want to see him in person. I’ve been told, when I was a kid, that dream about anything and it comes true in your life because universe is there to grant your wishes. And I was like, I want to see this person in person, I don’t know how and suddenly I get this huge advertisement saying that, hey, Kirill’s coming to San Diego, SuperDataScience GO, and I’m like, I am getting the ticket right now. First night I landed in San Diego, went to the hotel poolside, Kirill was there and the moment I saw Kirill, Kirill saw me and we recognized each other and we hugged, and it was just an amazing moment.

Kirill: Yeah, it was love from first sight.

Neelabh: That’s what I told my colleague and he was kind of jealous, but I was like, I’m enjoying this moment. It was amazing.

Kirill: Yeah, it’s funny because as you say this, in my hand I’m holding a pen from the Data Science GO event, it was such a blast. I don’t want to blow my own trumpet or paint a picture, but I personally thought we were doing it as more of an experiment, can we pull it off or not. I thought it was really great. What did you think of the event?

Neelabh: It was. Indeed, it was such an amazing event. You call it experiment, I call it the super hit event. That’s what I call it.

Kirill: Thanks. It’s all the dancing. Because you definitely weren’t expecting the dancing and the motivational speakers there, were you?

Neelabh: Oh man. My favourite part of all the motivational speakers was Laser Sharp Commitment from Ben Taylor and Being Compassionate by Urie Suhr. The entire ambiance was so positive, everybody just wanted to learn more, grow more and just be amazing. That’s what the entire ambience of that hotel was. The dancing really got my blood flowing and it was just amazing. I cannot stop saying it because it was just amazing. I’m waiting for 2018 SuperDataScience GO.

Kirill: Yeah, me too. Can’t wait, and that’s going to be epic. Anyway, getting side-tracked. We’re here to talk about you, about your career and the amazing turns and twists that you’ve had in the path that has brought you to data science. Are you ready for this?

Neelabh: Oh, absolutely. Yes.

Kirill: Awesome. I read about your bio a little bit and also checked out your LinkedIn profile- very, very impressive. You started off in a Bachelor of Computer Science in India, and then you moved to Texas. Let’s go from there.

Neelabh: All right. Back then, I was just another teenager who loved technology but at the same time, I was struggling so hard with keeping up with all the technology, programming languages, algorithms and data structure. Somehow, I made good grades but at the same time I wasn’t, for some reason, enjoying it, probably because of the fact that back then I didn’t have much of resources. Whatever resources we had in our school, it was probably more than enough because then I realized that I really want to do some great thing in the field of computer science and technology. By the time I completed my bachelor’s, I wasn’t very much happy with whatever I learned in my bachelor’s. To be honest, it wasn’t enough. It was, according to my standards, it was nothing at all. I really wanted to do something extra and then I planned to move to United States, I wrote GRE. As a matter of fact, I was the first person in my college to write GRE, so now you can imagine.

Kirill: What is GRE?

Neelabh: It’s Graduate Record Examination. It’s a standardized test for international students or even students in United States who want to pursue a graduate programme in computer science or STEM, which is science, technology, engineering and mathematics. I was the first guy from my university to write GRE. I scored well, I was happy about that, and then I started applying to different universities and I was fortunate enough for UTA to accept me; University of Texas in Arlington is where I did my master’s from. Then in 2014, January, I flew to Texas and then I started master’s from there.

Kirill: Nice, very nice. Bachelor of Computer Science, Master of Computer Science and what was the difference? Did you feel that in your master’s now you were getting what you were after?

Neelabh: Absolutely. It was just a 180-degrees change from what I studied back in India and what I was experiencing then in my master’s. It was just amazing, especially the resources that you have in United States for any STEM student, it’s just mind-blowing. You’ve got so much of resources that you cannot … Don’t take me wrong but sometimes you don’t even have to go to classes if you really want to study on your own because there are so many online resources, digital libraries and people like you providing all those courses and stuff. Going to the classes is just cherry on top because then you actually get something really professional. I started my career with spatial databases. I was a thesis student in my master’s and I really got a lot of interest in spatial databases, dealing with maps and geolocation services and stuff like that. I got a really huge interest in this field and especially in the indexing part of the spatial database. I was like, this sounds really cool. I really studied hard for that; days and nights and finally did my thesis on comparison of different spatial indexes and which one is the most efficient in certain conditions. In databases, especially in spatial databases, you’ve got different query types, so for a specific query, this index is better, for another query, this index is better. Or maybe the indexing system, which is really amazing, has its own trade-off where it takes a lot of time when it’s getting entire data and making the index on the entire data set. That was a very extensive research that also got published as a research paper, and that’s what my master’s thesis was all about.

Kirill: Before you go continue, can you get us up to speed because I think maybe some people are lost, including in me. What do you mean when you say, “indexing of spatial databases”? Can you give us an example or something to just understand that concept there?

Neelabh: Sure. If you talk about a database, we solve all these databases like B-trees, B+ trees and the regular indexes that hold some values. The regular data types, the integers or strings, or stuff like that. Spatial indexes, I will give you two examples. One is the R-tree it’s known as Rectangle Tree. Another one is the newer version of R-tree which is R*tree which is Rectangle Star Tree. All these indexes do not just store the characters or the strings on integer objects, but they also store objects like spatial objects, like points, lines, and polygons. Whenever you’re trying to index something like how far is New Jersey from Manhattan. Such kind of indexes have polygons of New Jersey, polygons of New York State, stored in their own indexes, and then they do the distance algorithm like Dijkstra's algorithm or stuff like that. But the access time by using all these indexes, gets so much faster that the computation of such higher spatial objects becomes really faster, optimized and efficient. Just to access much faster all these spatial objects, we use spatial indexing system instead of using the regular B-trees or B+ trees which are highly inefficient when you’re talking about points, lines and objects, there we have to use spatial indexes like the R-trees, R*trees, Quadtrees, and there are many more others.

Kirill: Tell me if I’m getting this right. Let’s say I have a map, it has all these polygons, all these lines and points. One way of indexing is like go through every single object and just give it a number, first object, second, but that’s going to be inefficient. You guys come up with spatial indexing systems that say the big polygons get numbers 1-10, the smaller polygons inside the big polygons get numbers 10-100, then these lines all get these… That’s just like a very basic example, but it’s a smart way of numerating the objects so that you can navigate through them better and run your algorithms and queries much faster.

Neelabh: Exactly. You’re absolutely right. Let’s say you’ve got this two-dimensional plane over which you’ve got the entire geography, what R-trees or R*trees do is they put the rectangle on top of it. By going with the algorithm and by going with the hierarchy, they start putting more rectangles over a specific geography so that we know how much granular a certain rectangle can access the data from this specific geography. That rectangle will only have, say, north-west of geography of the entire map, another rectangle will have the south-west of the entire map and now we actually know which rectangle is holding which geography over that two-dimensional plane, so that becomes more efficient to access all those geographies.

Kirill: Very, very, interesting. I didn’t know this actually was the thing until now and I will pose a question to the audience. Ladies and gentlemen listening to this, just ask yourself, have you ever used Google Maps? Probably everybody has, and we just take it for granted all the time, but I’m pretty sure they have something similar going on in there. What would you say?

Neelabh: Absolutely. I wouldn’t say exactly which indexing system because I’m sure Google would have come up with a more scalable approach because you already know there are billions of users who are using Google Maps every second of the day. But I’m sure these types of technology, other ones that all these routing companies are geospatial companies. As a matter of fact, the biggest one which is Esri, Environmental Science Research Institute, which is the creator of GIS, that is the biggest GIS company in the entire world, their headquarters is in Redland, California. These guys use indexing like R-trees, R*trees. There’s another one, GiST, it’s known as Generalized Indexing Search Tree, which uses a combination of R-trees and R*trees. All these huge geographical information system companies are using these indexings and as a matter of fact these indexings are also not new. Some of them came out in late ‘80s or early ‘90s so they’re fairly old and they have been extensively used since a long time.

Kirill: Gotcha. I’ve actually used Esri myself on an engagement once. When I was working at Deloitte, they preferred a different one. Do you know Pitney Bowes? We used them for geocoding but then then the program that we were using was called Pathfinder, I think, I’m not sure. I found it very interesting to work with georized data because you get very cool insights. For example one of the projects I was working on is when you need to estimate drive time from one location to another, instead of just doing a circle around your location … Or what is your customer’s catchment? Instead of doing a circle around your store, you actually use the roads because sometimes roads help you go faster, sometimes roads make you go slower, and based on that catchment, drive time catchment is different to just a circle around your location. That helps build your customer profiles better.

Neelabh: Absolutely.

Kirill: Awesome. Good to hear some insights into how these indexing systems work. Is that what you’re doing your PhD on at the moment?

Neelabh: No. Right after master’s, I was accepted as a PhD student under Dr Ramez Elmasri. If somebody loves to read about databases, Dr Elmasri is one of the big guys in the field of databases. I was fortunate enough to be a student under him and then shortly after that, I started doing my PhD back in ’15, again in indexing systems but a funny story, that semester, I wasn’t getting any courses at the university. As a PhD student, you still have to take a couple of courses and the only open ones were machine learning and neural networks. As a matter of fact, not neural networks but then machine learning was the one which was open. I enrolled myself in that course and ever since then, I did not look behind. I started machine learning days and nights and I told myself that I’m probably going to get married to machine learning. In 2015, I was also introduced with Dr Andrew Ng through his course on machine learning, I think it was on Coursera. I was introduced with Dr Andrew Ng and that course really opened my horizons, it really broadened my mindset regarding data science and how can machine think and what kind of cool stuff that machine can do in today’s time. The biggest takeaway from that course was I really got good in Octave and Matlab. I actually started coding. I started coding in C before, but ever since I got introduced with Octave and Matlab through Dr Ng’s course, I was like, wow, machine learning is just so cool, it’s just probably the best that could ever happen to a computer science student in today’s time.

Kirill: Nice. But why did you think that? There’s lots of people out there who haven’t tried machine learning, what would you say to them? Why did you think it was the best thing that could happen?

Neelabh: To be honest, Kirill, I think this was probably the only thing in my life in computer science that was tangible to me. This is something that I could feel from my heart and everything made sense in machine learning. Whatever I gave to the machine, whatever I was expecting it to do, it was coming out of the machine and in such a logical sense. Everything made sense. Besides that, I was able to see the future and that’s what I have always dreamt ever since I was a kid. I was able to see what’s going to happen next before even that thing happened. That was the thing that really got my attention. I was like, wow, I really can predict something even though I don’t know anything about it? That’s what machine learning really taught me. I would definitely say that that’s probably the best thing that could have ever happened to me.

Kirill: That’s so cool and it’s great to hear the passion in your voice. You’re speaking, and I can feel how passionate you are about this. It’s even more exciting that this random chain of events happened that you signed up to this machine learning course and you discovered it. How crazy is that? Like you say, you ask the universe for things, it will give them to you. And here it’s just like: by the way, you’ve got to be passionate about this, how about you sign up for this course?

Neelabh: Like I said, this was the only open course. But at the same time, I did not have any other option, I had to go in the course. It was like universe is calling me to be a part of this data science community. It was there for me, it was just waiting for me to get in there and realize what potential I have in this field. There you go, and now we’re talking to each other.

Kirill: Exactly. That’s crazy, that’s crazy, man. Tell us a bit more about your PhD. Without going into too much technical details, like at a high level.

Neelabh: What my PhD is all about, is about seeing the future. I’m trying to see where a user is going to be next in his future time. I’m trying to predict the future locations of a user based on his historical patterns which is coming out of his GPS data. I’m trying to see … until now this user has been here, here, here, and if you know humans on a typical day follow the same routine, no matter what. He might take days off and go for vacation or whatever, that can be considered as outliers, but on a regular day, he will still perform a regular behaviour or pattern, what he has, or she has been doing since a past couple of years. But at the same time, again, it changes over a period of time, so I’ll give you an example. As a student at a university, you are enrolled in some specific number of classes for a semester, so you are going to different buildings or building A, B, and C in this semester. Now the semester is over, and you enrol to different classes, now the buildings that you go are D, E, and F, so suddenly, within six months your entire pattern has been changed. I’m trying to make a machine or a model that can intelligently understand the pattern of this user such that it can predict where this user can be in the next given position or the next given time. The applications of this model are just out of my mind, there are so many applications. One of my favourite applications could be recommender systems. All these different companies can advertise their products based on where the user is going to be next. Let’s say a company like Walmart, you’ve got a Walmart app on your phone and then the Walmart has been recording your data, of course they have to take your permission because that’s a really sensitive data. But if you happen to give your locations to Walmart, Walmart can make use of this model and can predict, looks like tomorrow at 12 pm, this user is going to cross Walmart at this street, let’s just start giving all these recommendations or advertisements or maybe coupons for this user just to come in while he’s going to some place. Because he’s going to cross Walmart, that’s what our model tells and it’s a win-win situation. The user gets coupons, Walmart makes more revenue. That’s one of the applications that comes to my mind right now.

Kirill: That’s mind-blowing. When you just said it, predicting where a person will be, that was kind of cool. But when you put it into context of how you can apply it to a business situation, that’s just crazy. It’s on the verge of sci-fi type or even … you know what I mean. It’s kind of very invasive even, like you’re living your life and all of a sudden you have this coupon for a store that you’re just passing by, that’s crazy, man.

Neelabh: Yeah. There are so many applications that me and my professor were talking about.

Kirill: Can you give us another one as an example, in addition to this Walmart one?

Neelabh: Another one that we came up with was about the medical insurance companies. Insurance company is kind of a business where they just want money but at the same time they try to not pay as much as it is required.

Kirill: Yeah that’s their business, right. They’ve got to balance out the risk and the revenue, it makes sense.

Neelabh: Exactly. So, let’s say all these insurance companies whosoever it is and if they have their customers, they ask their customers to give their GPS data to them. It’s going to be totally disclosed from public appearances or whatever kind of contract they’re making with their customers, just to make sure that their security and safety is not breached. Once they have the GPS data, now they know where the user is going to be next at every second of the day or every hour of the day, so on so forth. Let’s say they recognise, let’s say there was this hurricane in Texas. Hurricane Harvey and then Hurricane Irma which really hit the coastal side of Texas. They already have predicted that the hurricane is going to hit that place but at the same time a user or their model has also predicted that their customer is going to be somewhere around that area. Now look at this. Be it medical insurance or be it car insurance, if both of the companies have similar data of the user, the car company can tell the user not to go there because it’s kind of possibly risky. Health insurance can tell the customer not to go there because it’s possibly risky. They’re telling this information to the user a month or two months prior to the event, because they have already predicted where the user is going to be. Even if the user is not going to be at the similar location, they can also predict the entire trajectory that the user is going to be traveling so they can probably ask to take another route because this trajectory is probably broken or there has been an accident taking place. And that is happening way before the event actually took place. So, when you can start predicting the future, you can actually make a lot of amazing business movements that can actually harness the power of predictions, make more revenues, save lives, give coupons, make recommendations, whatever comes to your mind. This is just one model out of which you can think about millions of possible applications.

Kirill: That’s really cool, man. I love that description and like saving lives, that’s amazing. I wish this would be implemented very soon so that we can start getting the benefits of it. Also, you mentioned on your LinkedIn that you’re studying deep learning methods. Are they also integrated into all of this that you’re doing?

Neelabh: We started with traditional machine learning by making use of Markov models and hidden Markov models because those are essentially the time series modelling. Then we realized that hidden Markov models are really good when it comes to speech recognition or text recognition because those elements like speech or text, they are bounded. What I mean by bounded is if you’re talking about English, there is a certain boundary around English language that can be utilized in your daily conversation. But talk about human behaviour, human is an open system and the function or the movements, the style of travel behaviour the human possesses, that function is so complex that you cannot ever estimate. Once you estimate that function, it gets so easy for you to predict the future values. We then realized that hidden Markov models wouldn’t be the best model, we should also start looking into advanced predictive models and then that was a time when I enrolled myself in neural networks class at my university. Man, the moment I got into that class from the first day onwards, I was in love with neural networks. I was like, this is the game changer, and this is something that can really be implemented in my research and I can do so much cool stuff. Not just predicting locations, but so many stuff, so many function estimations that can only be dreamt before. Ever since then, my life totally changed. I even started talking functions, I started talking to numbers, I started talking to matrices. It was more like all these things, all these statistical models and algorithms, are just around and I’m playing with all of those. And so much so that I could use up all these statistical models in my research.

Kirill: When we were preparing for the podcast, you mentioned an interest fact about yourself. Do you mind sharing?

Neelabh: That’s a very funny fact. Unfortunately, or fortunately, this did not happen with me once, but it keeps on happening to me in recurring fashion. I’ll tell you my experience when I first experienced it. I went to bed after my day of research and studies. I slept and somewhere about four in the morning or five in the morning I woke up and what I dreamt was I am a cell inside a matrix and I hold a value, and my function, my job within that matrix is to get the value threshold computed and forwarded to the next following matrix. That day I cannot ever forget. That morning I told myself I have given myself to data science because I know that there is a reason that I’m in this world and the reason is to fulfil the purpose of data science. And I’m glad to be a part of it. That was crazy.

Kirill: That’s awesome. It’s definitely a whole next level of data science when you do it in your dreams, that’s really cool. Okay, so you’re doing your PhD, when are you scheduled to complete it?

Neelabh: I completed my proposal this semester. A proposal is the third stage out of four stages in PhD, and I am planning to graduate next semester. As a matter of fact, I also got a full-time offer as a data scientist from Walmart so I’m talking on as a ….

Kirill: Not surprised, not surprised at all. [Laughs] With your coupon idea.

[Laughter]

Neelabh: During my interview, this guy was like, this was my final stage of the interview. The interview went for almost a month and a half. At my final stage of the interview, the director of data science asked me, how can you leverage the power of your research in our community. And I never even thought about it, it was just then and there that I came up with this idea and he was like, okay, whatever, even if you don’t get to work with us, you are connected to me. So just be in touch because we can do stuff together. I was like, man, this sounds really cool, why didn’t I think about it before? But yeah, probably next semester, possibly from the month of May I will start working with Walmart.

Kirill: Nice. Congratulations, that’s awesome.

Neelabh: Thank you. Thank you so much.

Kirill: Very great to have an offer, a full-time offer lined up while you’re still studying. It’s definitely an accomplishment. What advice would you have for students listening to this podcast who are soon to graduate? What’s the best way to go about lining up a job for themselves even before they finish university?

Neelabh: I would definitely recommend to plan way, well before time. You cannot just leave things lined up for the last moment. Start applying as much as you can but do make a list of your 10 most favourite companies that you really want to work for. Make your resumes, tailor your resume according to those 10 and forward it to them, and just hope that you’ll probably hear something back from them. But even if you don’t and if you’re getting rejections, let me help you here. I got 150 rejections, Kirill, before even I got my interviews from a couple of companies. In your life, don’t be surprised if you’re getting rejections because that’s part of the game; but make sure whatever you are doing, give your 120% laser-sharp commitment. Get married to whatever you’re studying because then and there you will know that you can win it. Don’t ever lose hope. Believe in yourself and you are going to get it because you are worth it. Why else are you working so hard on yourself? And I’m sure, I believe, if one is 2000% committed to his job, nobody can stop him, and the universe is there to help you out. So, yeah, rejections, that’s a part of life. Learn from it and be better every day. You’ve got to beat yourself every day. That’s how I try to live my life. To be honest, Kirill, wherever I am, I know I haven’t achieved as much, but at the same time everyday I’m trying to grow because I know if I am trying, I’m going to make it. That’s what I want to tell all these people, whoever is listening to this podcast.

Kirill: Love it, man, love it. You should be a motivational speaker. I felt you’re honest. That was so good. That was really touching and I’m sure you’ve helped a lot of people just by that message and I totally agree with that. Rejection is part of it. At the same time, I’m sure there’s already a ton of companies who didn’t invite you for interviews that would have they known what you’re doing for your PhD and would have they known what exactly, how you can help them with these great ideas, right now they’re probably kicking themselves in the butt. Saying that, damn it, we should have gotten Neelabh on our teams. At the same time, you have a part-time job right now. Is that correct?

Neelabh: That’s right. Currently I’m working as an intern. This company is known as Metal Roofs of Texas. This is a house improvement company, they help people to install roofs, floors, glassworks and stuff like that but at the same time within this company, the owner, Mr Josey Parks, he is amazing. He is a 29-year old entrepreneur, very successful person but at the same time learning every day. He thought of changing this contractor business altogether and opened a new company within the parent company, that is Metal Roofs of Texas. This company is known as Cognitive Contractors. What we do is, we are more like consultants. We try to get other companies’ data and we try to help them in their analysis because in the field of contractors, Kirill I’m telling you, today people are so old-school and they’re still depending on zip codes. And they are so much convinced by the fact that okay, this specific zip code is the one where I can make most revenue. They are not thinking about anything else, they’re not getting demographics, they’re not getting house data, they’re not getting even people, incomes data, or whatever, you know. They are just considering that over the past five years I’ve made so many million dollars from this zip code and I’m not moving anywhere else. They’re just in this box and they’re not ready to think outside the box. But this guy, Josey, he has a vision saying that, I need to change the way this entire- it’s known as blue collar- business is working. He’s trying to get all these data which has been collected over the past 20-25 years, trying to analyse them and we are trying to target customers based on the revenue that they have generated for this company. We are prioritizing our work based on the customers who’re going to make us more revenues and we’re going to look for the customers later who are not making so much of revenue. Let’s say we’ve got these … on a typical day we think that according to our model and analysis, we think these five customers are going to make more than average revenue for us, but these guys are really spread across the city, so the biggest challenge for our salesman is to contact all these customers so that they can advertise the product and make more revenue. Again, another problem that we face is the traveling salesman problem where we have to make the most efficient route to contact all these customers, but again in the prioritized manner. Again, we go through with the entire analysis and we see which of the customers can be more prioritized because those are the ones which we really need to contact. And then again, now we know that the past behaviour of this customer is something … this customer is not available at this current time or at this specific day, so we really have to take care of different dimensions altogether such that we are making the most intelligent decision by also saving more time and by using less resources.

 My sole job there is right now ETL which is probably 70- 75% of my entire time in data science. Thanks to you, I really learned a lot from your data science course with the SSIS and SQL-server that is really playing. That is probably the NVP in my entire ETL process. Once I’m through with the ETL, then visualization and my favourite is Tableau, again thanks to you. You’ve been an amazing person, Kirill, you’ve taught me so much. Tableau has been such an amazing tool. Initially I used to use – don’t laugh at me- I used to use Seaborn, [Inaudible: 39:06] and Matplotlib.

Kirill: That’s not a bad combination. That isn’t excel for sure.

Neelabh: Yeah but again, drag and drop and making your dashboards within like two minutes, that is a game changer. With the usage of Tableau and [Inaudible: 00:39:24] and Python, my work it’s really easy making all this analysis. And, interesting fact, sometimes I don’t even have to use machine learning and neural networks to make any sort of predictions because Tableau shows me some sort of visualization through which I can see that, okay, this visualization is following this specific function, and based on that, we can simply make the predictions. We don’t even have to jump into the complexities of machine learning. As a matter of fact, I saw your video where you were presenting, and I still remember you had … exactly. I was like, wow, this makes a lot of sense and that’s what I personally experienced in my job. And I was like, man. If you’re into data science, even though you don’t want to get into visualization, I understand but again have some basic knowledge of Tableau, it’s going to really help you no matter what.

Kirill: It helps with that data mining part. It helps you see your data and get those insights quicker. Like you say, you might not even need to apply machine learning if you can see that this looks like a logarithmic distribution or this looks like a normal distribution. It just makes things easier. And congrats on getting … Because I know exactly in which part of that data science A-Z course that [inaudible: 40:51] distribution presentation is, of me presenting. And the fact that you’ve seen that video means you’ve got to that part which is at the very end. I watch these statistics, I monitor the statistics on how people are going through the courses. Unfortunately, a lot of people drop off but like that is like a surprise lecture at the end of the Data Science A-Z course and the fact that you got to it, that’s really inspiring. Good stuff, man.

Neelabh: I cannot even believe that. Why would some people drop it especially when you are teaching? That’s like, what are you doing with your life? Come on, it’s Kirill, you’ve got to listen to him.

Kirill: Thanks. Appreciated. All right so that’s really cool. It sounds like this owner of the company you say is very young, 29, and he’s really changing the way this industry looks. That’s a very cool way to disrupt the industry. Have you noticed any positive effects of this? Has the company noticed how the sales have increased or the customers are happier and things like that?

Neelabh: Yeah. This guy is just amazing. This guy has such a zeal and passion towards data and he’s the one who appreciates the data more than anybody I’ve seen around my community. He knows how to tackle it. The biggest problem right now with me is I come from a computer science background and computer science has taught me things like dealing with different tools, ETL, cleaning, stuff like that. So, everything that I’m talking about in computer science is always technical. I am not so good when it comes to finances and money, but this guy knows how to deal with finances and money and even to convince a person to give his money to the company such that we can offer him our products. With our analysis, we have seen so much of difference in our own home company. I actually started analysing Metal Roofs of Texas data. We saw a difference in our sales and how sales people started behaving in a way that the analysis was asking them to do. Before, the sales people were just going randomly to different people without knowing or without analysing all these customers available and what time of the day they are available. That is just one small example. After doing an extensive research, I cannot go much into further details but after the analysis, the sales people actually changed their mindset. They know how data mining and analysis can actually change the behaviour they have been contacting customers. So much so that how can they even talk to them. Because the way they were talking before, we analysed the text data by making use of stuff like Wordcloud and stuff. We also analysed which were the words which were most occurring in the conversation with the customers. Unfortunately, those words were not playing a good role because some words are good in some context but at the same time you shouldn’t use them. We realized that text data analysis was the biggest game changer in the analysis in our home company, and how sales people should communicate with the customers. We actually saw a growth of almost 1.4 times. That was really a game changer and I am really pleased that while I was there, working … I’m still working with them, sure, but while I was physically there, I was able to see the change in the mindset of people who are not data driven before but now, again, they cannot step outside without seeing the data first. That was really amazing to see while I was there.

Kirill: That’s really cool. That’s a great example for the entrepreneurs and directors and executives listening out there. A great example of disrupting your own business. I think Richard Branson does this in his companies that he says … or I’m not sure who exactly, but let’s say Richard Branson. He says like, okay, how can you come and create a business that will put me out of business. He tells his own employees that. How can you come up with an idea that will disrupt my own business? And then, once they do come up, you just adopt that idea. That’s exactly what you’re doing here. You’re like, okay, how can we approach this in a different way that if we were competing with ourselves, we would put ourselves out of business. And then you just go like, 1.4 times the operations revenue, customer satisfaction, whatever, is a great competition with yourself and so you just put it into action and that’s a really cool approach and hats off to you guys for that. That was awesome.

Neelabh: Yeah. We’re just trying right now. I’m sure we’re going to be good with the kind of business that we’re trying to get into with the contractors’ universe.

Kirill: Yeah. For sure. I wanted to ask you. You mentioned that after you did all these implementations, the sales people … Before they didn’t think about using data and now they can’t even imagine stepping outside or starting their work without getting the insights from the data. Did you experience any road blocks, or did you experience any pushback at the start? Was it hard to integrate this culture into the company?

Neelabh: Yes. To a certain extent, yes. Like I said, you cannot just show a new technology to a person and just convince him. The person yet hasn’t seen what this technology can do because he cannot just believe whatever you’re showing them, so you have to give some examples before you convince another person that, okay, this technology is meant for you. Now, like I told you in this business, people are still not so modern because they are really convinced by the fact that the way this business has been going since years is probably the best way. We had to try because that was an experiment phase. We had to try first, do some analysis and let’s just do an experiment on a typical day. The sales people saw a difference and after that they started taking more interest in their sales. We were talking about how this can change the pattern or the flow of the information throughout the entire sales community and the way they were approaching customers. Initially it was a little challenging, but I knew that once they start seeing the difference … See, again, human nature. Humans are such an open system that it’s not hard to convince them, but it takes some effort to convince them first. That part was the time when I was trying to experiment with different analysis and fortunately it worked. The initial phase is always challenging.

 

Kirill: Yeah, I totally agree. I think a lot of people … Why I asked this question is I think a lot of people have or will in the future come across this problem that when you have people, employees or staff who are a bit more old-fashioned, a bit set in their ways, have never seen this new approach, you will have some kind of challenges. Because people usually are very resisting to change, they prefer to keep things the way they’re used to and the way that they know how everything works.

Neelabh: Yeah. It just takes some effort. The initial phase even in any predictive modelling it says you still have to show some data, that’s why supervised learning is a little better than unsupervised learning. Because you try to show the data first and let your machine learn and make some sort of predictions. And that’s what is with the humans because, I know, I mean, humans are reinforced but at the same time if we can show them something that can be helpful to them, it actually makes more sense to them after they have experienced it before.

Kirill: All these humans, they make me laugh every time.

[Laughter]

Sometimes on this podcast we get carried away talking about humans, it feels like we’re not humans, it’s funny.

[Laughter]

And I love your … this is a great quote. Humans, of course they’re reinforced, but sometimes they need something to show. It’s just crazy, right. Humans are reinforced.

Okay, I got an interesting question for you. You’re very passionate and very well-versed in lots of topics. Very diverse topics as well from GIS spatial indexing to machine learning, neural network, also I saw in your LinkedIn, genetic algorithm which we haven’t talked about. That’s a whole different can of worms, right?

[Laughter]

But before we jump into that, if we even do, I wanted to ask you, is there anything else on the horizon for you that you’re passionate about learning, that you can’t wait to get your hands on?

Neelabh: Right now, I’m still sticking myself and making myself better in whatever I am doing right now because next semester I have to graduate with a PhD so there are a couple of tasks which were told to me in my proposal that I have to present before I defend for the finals. So, right now I am really getting down into Tableau. I have already taken the advanced course of yours in Tableau and I am brushing my skills more in Tableau. But when you talk about neural networks, I’m really intrigued with the time series modelling in neural networks, especially the recurring neural networks. I recently started learning more and more about the long short term memory neural networks which are again time series neural networks and I’ve seen the way it can be implemented to predict the time series functions. We all are surrounded with time series functions like stocks, price exchange, even human locations. The next move is to study more about the advances in the LSTM. And there is one extra thing before I’m going to take a break and that is the Capsule network if you’ve heard about that.

Kirill: Yeah, I’ve heard about that. That’s a next step, yeah?

Neelabh: That just blew mind my mind. That is just amazing. To be honest, I never thought about it. I sometimes think about the advances and how can things be changed but this thing never came to my mind that can we show a three-dimensional image to a neural network so that it can understand it better? One of the examples that the creator, Dr Geoffrey Hinton, one of the examples that he gives is if you show a coffee mug to a neural network kept on a table but if you invert the image of the coffee mug, probably the neural network won’t be able to recognize the coffee mug because now it has been inverted upside down. I was like, that’s amazing. Once you start showing the images in three dimensions, now the entire number of times the dimension has been increased is making more sense to the network to understand the function. That is something I really want to jump in and to understand it.

Kirill: Yeah. Awesome. You know I asked you that question, it was a tricky question, but I felt there had to be something. A person as passionate as you, has to have things that they’re always waiting to learn. I don’t think you’ll ever have, like you say, a break. But you’ll always have something in your mind that you’re looking forward to. That’s really cool. RNNs and Capsule networks. I just learned about Capsule networks, Hadelin told me about them a few weeks ago, or a week ago or so. Yeah, it’s just mind-blowing with the 3D images, that’s something for people to look into. And then once you master them, which I don’t think is going to be long, you should come back on the podcast and we can talk about Capsule networks for an hour.

Neelabh: Sure.

[Laughter]

Kirill: Sure, no problem. Everything under control. I just have… We’ve got about 10 minutes. Okay, I’ve got couple of questions. Let’s do some rapid-fire questions, are you ready?

Neelabh: Yeah, absolutely.

Kirill: What’s the biggest challenge you ever had as a data scientist?

Neelabh: The biggest challenge was the data collection for my own personal research. GPS data.

Kirill: Did you get it off your phone?

Neelabh: Yes.

Kirill: What were you researching?

Neelabh: I wasn’t getting enough data for the predictions, to experiment on my models. This is a data set which is publicly available. It’s created by Microsoft Research Asia, it’s known as GeoLife. So this data set contains about 172 users from Beijing China. It’s an amazing data set. To be honest, it’s amazing, it’s mind-blowing. Some people have their historical locations saved for five years so that’s probably millions of rows of data. But at the same time, I already used that data set in my previous research but this time I really wanted something challenging, something I could touch, something I could feel. Then I was like, well, I’ve got a smart phone and I’ve got Google Maps in there. There should be some way for Apple and Google to collect my data. I looked into Apple and Apple does not let you have your own data. I’m not sure, but that’s what I read. And I was like, that’s not cool. Then I encountered somewhere that Google does it and all you need to do is go here, go here, go here … It’s a long process but eventually you get your entire historical data set. And I was like, cool. Let’s just work on this. But that actually took me more than a month to collect it, to clean it, to massage it, to make it perfectly ready. Because time stamps and stuff, you have to convert it in the format that you want it. The latitudes and longitudes are like probably up to eight digits, so I didn’t want that much, I just wanted to six digits. Basically, the ETL stuff. Then again, I was like just latitude, longitude and date, time and time stamp is not going to help you. I was like, what else is really important in a human’s life to make a decision of his travel. I was like weather is an important role. Just add weather to that. I had latitude, longitude, and I had date time stamps. I scraped the entire internet and I added the weather feature and believe you me, Kirill, I increased my accuracy by 1.6 or 1.7 times, just by adding the weather. That was amazing and that’s what my professor said. That that looks pretty cool. Yeah, that was the biggest challenge.

Kirill: That’s so cool. Just to clarify, how long was the … It wasn’t that for a month you were running around with your phone collecting the data. You already had the data in your phone, it just took you a month to get it out and massage it, right?

Neelabh: Right.

Kirill: And how long was the period of the data that you were working with in the end?

Neelabh: I had data starting February of 2014 until I did this early this year, so probably January or February this year. I had almost three years of data.

Kirill: That’s so cool. And so, did you find any surprising insights like, you know, how often you go to the bar or something like that?

 [Laughter]

Neelabh: That was probably the most visible place, yeah.

Kirill: Gotcha. Okay. All right, next question. What is a recent win you can share with us? Something really cool that you did in your role or research or studies.

Neelabh: Recent win. Like I was talking about the time series modelling, I was able to make a model that was generalized not just to predict the future locations but also was giving me so far accurate results when I was trying to predict the future currency exchange rates. It’s a short story. I was in the US a couple of months ago and I was like I sometimes have to send money to India and my parents sometimes send money to me. There’s an exchange of money from US to India and it all depends what the exchange rate is. If somehow, I can know when the rate is going to increase or decrease, I can make better decisions. I collected this probably past 30 years of exchange rate data in Indian Rupee and US dollar, and I made this model and was able to make a model that could predict whether the price is going to go high or low in tomorrow’s time. I also experimented with it and it was Friday evening when I was checking this model and it told me that the price was going to go high, which is a good thing because I was sending money from US to India and it actually went high by probably some couple of cents. It wasn’t a lot but again, the model did something good which helped me to make some more money. I also published this article through Stats and Bots on Medium. It actually went pretty good, I’ve already gotten almost 1,500 claps and people actually started noticing my work from there, started connecting to me. As a matter of fact, I also got an offer as a freelancer for this guy in Brazil and he wants me to make a similar kind of model for him. That’s again some extra cash that I can make from that freelancing job. That is probably my recent win.

Kirill: That’s so cool. That’ such a cool … Like you share a ton of wins and now all of a sudden, here’s another huge one. And I love how you took it to the next level, you wrote an article about it. That’s the way to go, that’s the way to get recognized. I’m just checking, is it called A Guide For Time Series Prediction Using Recurrent Neural Networks?

Neelabh: Yeah. There you go.

Kirill: Cool. We’ll share it on the show notes. Guys, check it out. It’s like very in detail. And also, especially people taking the deep learning course, definitely check it out, it might be something that can be very useful.

Neelabh: That’s the article where I’ve explained LST and the Long Short Term Memory pretty much in detail and shown the examples how by making use of graphs and using the time series model as the training said and the testing said. It’s pretty intensive and in depth. It will be a good read, I can promise you that.

Kirill: Awesome. Okay, and moving on to our, what’s supposed meant to be a rapid-fire of questions, what is your one most favourite thing about being a data scientist?

Neelabh: One most favourite thing. Just one? I’ve got so many.

[Laughter]

Okay. Being a detective. That’s what I would have been if I wouldn’t have been a data scientist. I love being a detective and with the help of data science, I can investigate data. I can see something which nobody has seen yet, just by looking at data as in a database or in an excel sheet. I can actually see what’s going inside there and I can actually gain some insightful knowledge just by looking at the data. This is one of my favourite things, but there are so many other things and I can talk to you for days and days about it. But, yes, this is one of the things that I love about it.

Kirill: Awesome. Love that answer and I love how you phrased the question. Remember at DSGO? Were you there for the panel discussion? Remember we had that question at the end, what would you do if you weren’t a data scientist? Remember what Ben Taylor said?

Neelabh: Ben Taylor said he would catch snakes, something like that.

Kirill: He said he would be a python breeder. Everybody is like, what just happened? Python breeder?

 [Laughter]

 And this is like 150 data scientists sitting in the room. Everybody is like, hold on. Is that Python the programming language or like an actual python, python? He’s like, no, no. A python. I would breed pythons. Those things go like $20,000 on the black market. Like, what is this guy? You’ve got to give it to Ben Taylor.

Neelabh: I still remember when he answered that, and that was the same thing I was thinking to myself. And I was like is that Python, Python, or python snake? That’s how I remember that he would breed snakes or something.

Kirill: He’s got some crazy ideas. He’s always great fun to talk to.

Neelabh: I saw his podcast, it was probably the video podcast that you had with him and he’s fasting and sleeping trends. That’s just crazy. Who does that? I mean, this guy is on another level, you know. He’s really serious in whatever he’s doing.

Kirill: His life has been crazy. He’s like slept in a tent outside on the campus in the winter for a whole semester and he does crazy snow skiing. One time I should get him on the podcast just to talk about the stories that he’s had, it’s crazy.

Neelabh: I know, yeah. That would be really cool.

Kirill: Anyway, moving on. Almost final question. From all the crazy things you’ve done in data science and machine learning, deep learning and so on, where do you think this field is going, and what should our listeners prepare for to be ready for the future that’s coming?

Neelabh: Before that, I just want to share something which will lead to my answer. There are so many crazy things that I’ve done with machine learning and the internet world, some things I cannot even remember but two things that I can remember. One is I used to be a GTA, being a TA for the class. Sometimes there used to be times when I wouldn’t be able to attend classes because I was busy on something which was really important related to my research. But then again, I really had to be in the class because I had to grade students based on their presentations.

Kirill: What is a GTA?

Neelabh: It’s a graduate teaching assistant.

Kirill: Ah! I was thinking grand theft auto. The computer …

 [Laughter]

 I was like, what? Because Hadelin keeps talking about GTA and we were about to create like a deep learning neural network which plays GTA. And I’m like, what is he talking about? Sorry, don’t mind me. Keep going. So, you were a tutor, right?

Neelabh: Yeah. I was a GTA who used to grade but not play grand theft auto. I had to be in the class along with my professor so that we both can grade. There was a time when I wasn’t in the class and for a student there was no grade from my side but there were grades from my professor, so we were discussing that; how can we solve this? Within like five minutes, I had collected all the data because I had been GTA for this professor since past two years now. We had two years of data within like 10 minutes, I made a model that can take my professor’s value and can predict my values. Whatever professor graded these students, I also graded these students. It was more like a linear regression. I gave the training data as my professor’s data and output as my scoring. For the testing, I gave the professor’s data on a specific given day, the student, all the information about the student and predicted my score. So, I gave this model to my professor and be like, hey, whenever I’m not in class you can predict what score I’m going to give.

 [Laughter]

Kirill: I love it. That’s so cool.

   [Laughter]

Neelabh: Yeah. I loved that one. And another one that I had in my mind, but my friend implemented this was … In India, you know how we have arranged marriage system where parents look for a girl, you know they try…

Kirill: I love where this is going already.

Neelabh: So, this guy was really fed up with the fact that his parents are like trying to hook him up with girls and stuff. So they already arranged all these tens of numbers of girls for him but he already knew these girls from before so he slightly had an idea. And that idea was converted to a score, 0-10. We took the entire data set, rather he took the entire data set of all these girls and trained the model and was looking for a model to predict the best match for this guy with a girl whom he hasn’t spoken with yet. Based on her different attributes. As a matter of fact, he got a score of 85 for this specific girl and he came back to India, met her, and I think he’s already engaged with her.

Kirill: Whoa!

Neelabh: Yeah, that was something like mind-blowing. I was convinced that neural networks and machine learning is the future. It’s definitely going to be the future. And these examples that I just gave you are simple machine learning, you don’t even have to go to neural networks. But think about it. How much more complex functions you can estimate by using neural networks? That’s what the future is going towards. There will be a time. I remember you saying that the amount of data which is increasing, and which will be in the year 2020 or 2030, it’s going to be humongous, it’s going to be a kind of data where we need everybody to understand data and the science behind the data. So just like everybody knows how to use an iPhone or an iPad, everyone should know what is data science and how can it help you. I’m sure, in tomorrow’s time … Like Sophia, the artificial intelligence robot, she’s already there. United Emirates has already given her citizenship, so we already have robots like Sophia, so what do you think? Tomorrow, we’re going to be living with robots. They’re going to be sleeping with us, who knows? This world is going crazy. I don’t know. I’m sure this world is going to be an amazing place and what people are thinking that AI is going to damage everything, no. it’s going to be there because we created them. They’re going to be our friends and it’s just going to be an amazing world to live in. Just be ready, everybody start reading, start learning maths, start getting more interested in statistics because that’s going to be the future. That’s going to be the future.

Kirill: Fantastic, man. Love the answer, love the examples, both of them. Crazy with the tutoring. If I’m not here, just use this model, you’ll get my score. And of course, you know, your friend. Oh, man, I really hope there’s going to be a happy marriage. You can use that as a case study somewhere. Like when you write a book one day.

Neelabh: Yeah, maybe. Who knows?

Kirill: All right. Well, let’s start wrapping up. First of all, I want to thank you so much for coming and sharing all this wisdom with us and all these examples. Where can our listeners contact you, follow you, or find you, if they want to find out where your career goes or what crazy thing you’re going to be up to next?

Neelabh: One answer. LinkedIn. I’m there on LinkedIn, I want to say probably 18 hours a day. There, I’m following people, I may not be posting as much as you would expect because I’m there for 18 hours, but I’m there mostly to learn from other people. LinkedIn is a place where I can be found, it’s another place where I live basically. Besides that, you’ve got my email from the Medium post. I also have my Google account, if you just go to Google Scholar, type in my name, you’ll probably find a couple of my papers which are already published. You will probably find some more coming up. Google Scholar, LinkedIn and my email on the Medium.

Kirill: Awesome. And I saw you also share some stuff on GitHub, maybe people can …

Neelabh: Oh, yeah. Absolutely, yeah. My deep learning repository, machine learning repository, my spatial data repository, feel free. Data sets are there, models are there, explanations are there. Feel free to play with them. Correct me if I have done something wrong because I am going to learn every day. So tell me where I’m wrong, we all can make it work together. Please, I’m always off on new projects as a matter of fact, so hit me up whenever you think it’s going to be amazing to work with me. Let me know, we’ll work out something together. And yeah, I’m always there. I’m always there for the entire data science community, for this entire data science universe basically. Hit me up.

Kirill: Awesome. That sounds pretty cool. Actually, really cool. And I’m guessing probably you’re going to get a few … I think it will be really cool if people who are listening to this podcast, who are doing a research paper and need some expertise in some of the things that you said, they’d actually contact you. I think like you can definitely provide some really cool insights or assistance with that.

Neelabh: Sure

Kirill: One final question for you. Do you have a book that you could recommend to our listeners to help them better themselves sand become better at data science, or whatever in life?

Neelabh: I have a lot of books in my mind but the one that I would share is probably … Everybody should have if they don’t have it, just get it right now, it’s on Amazon. It’s known as Data Science From Scratch, it’s by Joel Grus, the publisher is O'Reilly. Amazing book. You’re going to see data science the way data science essentially is. No packages, nothing, you’ve got to make everything even back propagation from scratch. The other one, another favourite of mine, is Machine Learning in Python, it’s by Sebastian Raschka, again on Amazon. That book is just amazing. It uses the most important packages in Python to teach machine learning and different algorithms inside machine learning. He even touches a part of neural networks so that you can have a preliminary, the most basic examples of neural networks and how they work. The most important, if you don’t have it, get it right now, Data Science by Kirill Eremenko on Udemy or SuperDataScience. Trust me on this, that course changed my life. This is coming directly from my heart. Trust me, that course is probably the best for data scientists. Get it because he’s going to tell you, he’s going to cover the most important four stages of data science. So you better get it. The last one which is my all-time favourite, it’s Linear Algebra and Its Application, it’s just a mathematics book, it’s just all about mathematics. Matrices and linear algebra. It’s by David C. Lay. Get it. Whenever you are bored, tired, get a paper and a pen and open that book and start solving all those questions. Not because it’s going to … It’s going to improve, definitely, it’s going to make you amazing in linear algebra but at the same time, I think if you’re practicing mathematics on a daily basis, it actually improves your logical reasoning and the way you think. It actually makes you faster in your thought process. Those are four or five resources that I can provide you with right now.

Kirill: Thanks, man. Really appreciate it and thanks for the plugin as well.

Neelabh: Thank you. Thank you, Kirill.

Kirill: I’m just going to recap those. Data Science From Scratch by Joel Grus, Machine Learning in Python by Sebastian Raschka, of course the Data Science A-Z course by yours truly, and Linear Algebra and Its Applications by David C. Lay. It sounds like you’ve got a whole library just off the top of your head. That’s a lot of resources. Thanks, man, I think people will really appreciate.

Neelabh: Sure

Kirill: Okay, thanks again so much for coming on the show and sharing all these insights and knowledge. It was a fun chat. This has been like one of the longest podcasts, but I really don’t want to end it. We’ve had so much laughs here, I really don’t want to end it but time has come. Thanks, man so much for coming on this show. This has been amazing.

Neelabh: Thank you so much Kirill. One last thing. You are awesome, your entire group is awesome. Be awesome, keep people teaching, making them amazing. You are the one, to be honest, you told me that I am passionate in whatever I’m doing. Yes, I am, I try to be, but you actually told me how much important it is to be passionate. I remember from your data science A-Z, I remember from Tableau, you said, even if you’re going out to present, your passion should be reflected with your smile and your words, and I cannot forget that. You know you said that so thank you. I have really taken you seriously and I will request everybody to take you, your group, your courses seriously, because those are an amazing asset. I will really say an asset because that’s an amazing asset in today’s time when it comes to data science. So thank you. Thank you for making all of us awesome. Thank you so much.

Kirill: Thanks. Really appreciate it. All right. Well thanks again and have a good one. Can’t wait to see you at the next DataScience GO event.

Neelabh: Absolutely, Kirill. Thank you.

Kirill: All right, so there you have it. I’m still so full of energy after this episode. Thanks, so much guys for checking it out and sticking through to the end. I really hope you got value. I really hope that if anything, you got the passion. Like, if you could feel Neelabh’s passion translate through the speakers or through your earphones, and it’s crazy. I love episodes like this. We had Nick Cepeda on the podcast who gave us so much passion, and now we have Neelabh and we’ve had a couple more. And it’s just so contagious when you hear people talk about data science or machine learning or deep learning like that or just the things that they are able to do with what they’re studying or what they’re passionate about. Of course, my… there are so many cool things that happened on this episode like for example when Neelabh used machine learning, linear regression to predict what his scores would be. That was such a good… there should be a meme about that I feel. It’s such a good example of how to use data science to optimize your life, to make it more efficient. My personal favourite takeaways were of course his passion about the topic of data science, and the crazy applications. The applications that he even mentioned, that Neelabh even mentioned at the very start with when you’re crossing a street and you get a coupon because the company that for instance if it’s Walmart, knows your normal patterns, your habitual patterns of location or moving around the city and they can predict where you’re going to be so when we were talking about it and there was no example, it was just like geospatial to predict a person’s location. That was pretty cool. But when you put into context and you give an example like that, or with the insurance companies that can save people’s lives, or car insurance companies they can help people not get into accidents because of the storms and things like that they know of or other road conditions, that is really cool. That is data science in action.

 So there you go, that was Neelabh Pant. Make sure to follow Neelabh on his LinkedIn, you can find the URL to his LinkedIn. Alongside is the Medium article which he mentioned which looks fantastic and alongside all of the other resources that he talked about. You can find all of that at www.superdatscience.com/115. There you’ll also find the transcript for this episode and that’s it for today. Make sure to share this episode around. Any data scientist you know at any level, share it with them and get them pumped, get them excited about data science, about machine learning, about where the world is going. Give to them some of this energy that you got from this podcast. And I can’t wait to see you back here next time. Until then, happy analysing.

[Background music plays]