Kirill: 00:00:00
This is episode number 403 with Data Science Instructor Juan Gabriel Gomila.
Kirill: 00:00:12
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.
Kirill: 00:00:44
Welcome back to the SuperDataScience Podcast everybody, super pumped to have you back here on the show. Today, we’ve got a very special guest, Juan Gabriel Gomila, who is not just a data science instructor, he’s also an Associate Professor in the space of data science at the University in Mallorca, Spain. Juan Gabriel had an amazing story of his own in getting into data science. He started out by studying forensics, chemistry and medicine. And then moved into mathematics, moved into becoming a developer, moved into data science, and then also became an instructor and then associate professor. So very interesting story and you will hear about his story, it’ll probably inspire you to continue and explore your own story even further.
Kirill: 00:01:29
And also Juan Gabriel is very passionate about the gaming industry, he is a Unity Certified Instructor so he actually teaches companies how to properly use Unity products and make the most out of them. So he brings a unique combination to his career of data science and the gaming world. And in this podcast, you’ll actually hear a case study that he will share of data science being applied in the world of gaming, so if you’re passionate about gaming, you will find this interesting and you’ll see how maybe you can apply your data science skills in the world of gaming. Also, we’ll be talking about gamification in data science itself, you’ll see how you can apply principles of gamification in your data science career, whether it’s for yourself to learn better, faster, or whether it’s in your presentations to help engage your audience. We’ll talk about teaching data science and we’ll talk about the future of data science. All in all, we’ve got a vert exciting podcast coming up for you. Get ready to be energized and without further ado, I bring to you Juan Gabriel Gomila.
Kirill: 00:02:47
Welcome back to the SuperDataScience Podcast everybody, super pumped to have you back here on the show. And in today’s episode, we’ve got a special guest calling in from Mallorca, Spain, Juan Gabriel. Juan Gabriel, how are you?
Juan Gabriel: 00:03:00
Fine, thanks. Here in Mallorca as you say, enjoying the sun and confinement again probably.
Kirill: 00:03:06
Yeah, man. It’s been a year, right, since we spoke last? I mean since we saw each other in person at Udemy Live last year, yeah?
Juan Gabriel: 00:03:15
Yeah, it was in October or something like that.
Kirill: 00:03:19
August. It was in August.
Juan Gabriel: 00:03:20
No, October, it was in Berlin.
Kirill: 00:03:23
October, I thought it was in August.
Juan Gabriel: 00:03:24
It was in October, I remember it was already cold.
Kirill: 00:03:28
Okay. Well, it’s almost a year. Yeah, so how have you been?
Juan Gabriel: 00:03:36
Fine. I had the chance, you know I travel a lot for my work. From the week after we met over there, I had to fly to Los Angeles to give a training, then come back to Spain, flying back to Mexico, then all Europe around. But then on February, everything closed and from that moment on, I’ve been giving online just from my house. And it’s been really quite interesting, this change of having to pick a flight, two or three times a month, and being able to be at your own house for six months, right now.
Kirill: 00:04:15
Have you been able to get some value out of this and maybe recharge, look at your life differently, what’s been the biggest positive takeaway from this experience?
Juan Gabriel: 00:04:29
I think that the first thing I’ve been able to do is to enjoy my house. I mean, I re-modeled my own house but I was never there for two years, flying back and forth. And finally being able to stay at my own house, being able to see what I’ve build and what I am able to enjoy, you look things very different, you start to value a lot more your own life, you start to realize, hey, you’ve got a garden, you can plant your own crops, you can enjoy your pool. And that was something that I was not conscious about when I was flying back and forth around the globe.
Kirill: 00:05:08
Okay, gotcha, that’s very cool. Before coronavirus, why were you flying so much back and forth?
Juan Gabriel: 00:05:18
Well, I became a Unity Certified Instructor. Unity is this video game engine that people use all around the world to create video games right now for PlayStation or laptop, and I became a certified instructor after one of my travels to San Francisco, that’s where we met for the first time, if you remember on the Udemy Live 2018 it was.
Kirill: 00:05:43
Oh yeah, that was a long time ago, wow. 2017.
Juan Gabriel: 00:05:51
2018 I think.
Kirill: 00:05:55
No, 2018? Wait, I didn’t go to the 2018 one, I went to the one before that. I think 2017.
Juan Gabriel: 00:06:03
So 2017. One of those?
Kirill: 00:06:07
Anyways, a long time ago, yeah.
Juan Gabriel: 00:06:08
Yeah, three is a whole life right now. So after getting to this Udemy Live, I was invited to go to the Unity offices because I already had some process published on this video game engine. And since they saw what I was doing in this Spanish market, they told me, “Hey, you’re having more students learning Unity than we have on our website. You are doing a super cool job, why you don’t become a certified instructor and fly all around the world teaching at the different studios?” And that’s what I did, I became instructor, I passed some certification exam.
Juan Gabriel: 00:06:48
After that, one week after the other, I started flying around to the different companies, they are big video games studios that are launching the next big successes on the actual consoles that requested some kind of training in different aspects, could be optimization in video games, could be about artificial intelligence, about different mechanics to be implemented. And it’s been really fun because I’ve been able to know a lot of people from different backgrounds, from different places that all have the same mistakes, they all failed on the same stages and it’s really good to learn from that, to be able to anticipate the next trainings and keep learning myself and teaching the others.
Kirill: 00:07:32
Wow, you truly have such an interesting background, and also combination. I love how you carried your background in gaming into your profession as well as data science, in parallel, and you do data science, you do gaming design and you do them together. It’s crazy, it’s dream job I think.
Juan Gabriel: 00:07:55
Yeah, I remember when I started my first job here at the Spanish video game company, that I thought that that was the job of my life. I mean, playing video games, learning about that, recreating your own world, playing to be God, designing your own rules on that scenario and picking the randomness using machine learning algorithms, it was like the job of my dreams. And after I quit and started doing it at a big scale, doing consultancy for other studios or giving trainings on behalf of Unity, it’s been super cool because you see how this job is so big, it has so different aspects you can focus on, that you don’t know what you will have to do on the next one, due to the different aspects you can focus.
Kirill: 00:08:50
That’s incredible. I hope you’re enjoying this episode, we’ll get back to it after this quick break. And Confident Data Skills Edition Two is out. This is the second edition of the book I published in 2018. Some time has passed since then, a lot of things have changed in the space of artificial intelligence and data science. If you’re not familiar with the book, then it helps develop an understanding of all the main data science algorithms and the data science process on an intuitive level. So no code, no complex mathematics, just intuitive explanations of the algorithms and useful practical examples and case studies. This book will be extremely helpful for you if you’re starting out or if you’re looking to cement in that intuitive feeling for algorithms as your progress in your career. Specifically, you will learn about decision trees, random forests, K-nearest neighbors, Naive Bayes, logistic regression, k-means clustering, hierarchical clustering, reinforcement learning, upper confidence bound and Thompson sampling. And in this second edition, I also added robotic process automation, computer vision, natural language processing, reinforcement learning and deep learning, and neural networks.
Kirill: 00:10:03
Plus of course you will learn extremely valuable skills for a career such as ethics in AI, presentation skills, data science interview tips and much more. So if you want to get a grip and really cement in your intuitive understanding of this field, then this is the book for you. And you can get it on Amazon already today. It’s called Confident Data Skills Edition Two and it’s a purple book. So enjoy and let’s get back to the podcast.
Kirill: 00:10:31
So tell us. Let’s start from the beginning. How did your career go? I think this would be a really cool case study for our listeners in terms of absolutely from any path you can get into data science and yours was, as I understand, from a developer to a game designer to a data scientist to an instructor. So tell us a bit about that.
Juan Gabriel: 00:10:56
That is a really interesting history and in fact, I had the chance to [inaudible 00:11:04] a TEDx talk last year, because people were so interested about knowing about that. That the TEDx from here, from Spain, they contact me, “Hey, why don’t you speak about your life, [inaudible 00:11:18] your different stages, your experience, from learning to teaching and everything you’ve done in the middle.” And the TEDx really enjoyed that conference so in some little words.
Juan Gabriel: 00:11:32
When I was starting, when I was 15, 16, I wanted to become a doctor, I wanted to be a forensics doctor, I really enjoyed watching CSI on the TV and see how the detectives were able to investigate about the crime scene, about solving everything, and I thought that it would be a really nice job. But when I had to take the biology subject at the last year at the college, I had a really bad teacher. A teacher that came to the class, she just opened the book, stated reading page after page and you didn’t know anything because reading a book to an audience is really not the correct way of teaching. In that case, I quit that subject and I chose math implications so I just took some different subject about mathematics that I would be able to see at the University the next year. And I enjoyed it so much, about mathematics, then I decided to do a master’s degree. And it was a quick switch to biology, medicine, to mathematics, that was the first one, okay?
Juan Gabriel: 00:12:39
From then on, I started learning about mathematics, I finished my degree, I got an internship to go to Paris, but I had my girlfriend here in Mallorca, so I had to decide between going to Paris or staying here with my ex-girlfriend, right now. So I quit. I decided not to take that internship, just to let it go, come back to Mallorca and decided to look for anything to do. In that case, there was only one master level here on my island, it was the masters in teaching education, to become a teacher in that case. And that’s where I make the second switch, from mathematician to teacher, teaching about anything. But I saw…
Kirill: 00:13:35
That requires a lot of courage, completely changing your life for the second time.
Juan Gabriel: 00:13:40
Yes, because I was just 23 years old and this scholarship was super huge, they paid me for a whole year in France, in Paris. They paid for all the expenses I could have for getting into the University and all that stuff, everything was paid. But yeah, it takes some consideration to let it go and decide to come back to Mallorca. It was not an easy decision. But at that moment I thought the girl was worth it so that’s what I did back in 2011.
Kirill: 00:14:17
Love, yeah?
Juan Gabriel: 00:14:19
Yes. So I come back here, I do this masters in teaching education, and I see that education is still using a blackboard and a chalk and all that stuff instead of using new technologies. So I though okay, all the students have a smartphone, they use it a lot, it was when WhatsApp started being really [inaudible 00:14:43] between people of 12,13, 15 years old. And I thought, “Okay, why don’t I create a mobile application in order to learn?” So that’s what I did, I created small applications teaching about statistics, about the typical balls falling down and creating the bell curve, about solving a linear system equation, something like that. And the first day I go to class and I say to my students, “Hey, you can take out your smartphone and use that application to learn that.” They watch me with some weird eyes, “No, you want to take our smartphones, you are tricking us, that’s not what you want.” Yes, you can take that out.
Juan Gabriel: 00:15:31
So it was such a success, everybody learning with their smartphones instead of paper and a pencil. I realized that there was a lot of room for improvement both for education and integrate, so I decided to keep creating applications and some videos games, launch it to the app store, and I’ll have my first salary in that case. I was working from home, I was earning some money from my applications, but there were people that didn’t understand about that, and they were my parents. My parents were already around the 60s, they don’t have smartphone, and when they saw that I was working from the couch of my house, they thought, I don’t know, that I was doing some drugs or selling some bad things around the neighborhood to earn some money. So they told me, “Hey, do a curriculum, send the CV, look for some job, don’t work from the couch, that’s not legal,” or something like that.
Juan Gabriel: 00:16:30
So that’s what I did, I created a CV, there was a company in Mallorca that was starting, really famous, with just 12 people, it was creating video games and it was gaining some success. And that’s what I did, I went, just left a CV on the table, they called me for an interview the next day and they asked me, “How many years have you been doing that, have you been creating mobile applications and video games?” And I told them, “Years.” It’s been three months. You’re hired, the next day I was hired.
Kirill: 00:17:06
Wow, because they saw what you can do.
Juan Gabriel: 00:17:10
Yeah, they thought that, wow, you need a lot of background in order to do that. But no, you just need to investigate a lot to learn about new technology, to learn about new ways of applying things. Yeah, I have the background in mathematics for sure, and I already know a little bit about programming, about computer science. But it was a way of putting things together in order to be able to create these first projects I did. For sure, right now I see those and I see that I could have done better if I had the knowledge I have right now. But for the company, I was more than enough to hire this 24 years old boy to be one of the next developers of the company. But the inflection point comes here, because I was hired as a developer because I was creating mobile applications and video games. But the first day of my job, they saw that I was a mathematician and they tell me, “Hey, you’re a mathematician, we have a lot of data that we have been collecting for the last year or so, will you be able to investigate a little bit about this data and see if there’s margin for an improvement or doing anything to become more successful on that?”
Juan Gabriel: 00:18:36
And I told them, “Okay, let me run this.” Back in 2012, there weren’t the knowledge we have nowadays about machine learning algorithms and all that kind of stuff. Everybody was using Excel. Imagine yourself doing big data with just an Excel, you’re doing some linear regressions, even trying [inaudible 00:19:01] some data, replacing some missing data and all that stuff. And during three months, I invested a lot of time creating a dashboard that was able to get the data from the database, prepare everything to show to the bosses, and to the investors. And even looking some points where we could be better, for instance, creating clustering for our users, when you work on free-to-play video games, the biggest problem is how to convert a free user to a payer user. So over there, we did some segmentation, some clustering about the different kind of segments that we were able to find on the patterns of the game play of the players.
Juan Gabriel: 00:19:49
And the company was able to grow monetization 10 times, so if it was doing $1,000 per month, it was $10,000. After just three months, we were able to go from 12 people to more than 40 people. And nowadays, it’s still one of the biggest video game companies here in Spain. So you can how I switched from developer to a knowledge in just on company in that case.
Kirill: 00:20:20
Wow. And all that thanks to data science.
Juan Gabriel: 00:20:26
That is where I first saw what data science was, how I did have to know about the field I had to analyze. I had the mathematics, I had the computer science techniques, I know how to code, but I didn’t know about the field so I start investigating a lot about gaming. I start to see that all the companies don’t share their data, don’t share their findings, and that’s okay, because well, that’s the key for their success. But that was also the key for me to be able to teach that in the future, I saw that there was a lot of room for improvement.
Juan Gabriel: 00:21:06
That there was a lot of misknowledge about the techniques, about how to do things. And that’s where I started to create my first blog, it was really old in that case. But I started to post about how to analyze out layers, how to do some simple KPI’s arrangement, how to collect some data and transform into valuable data. And that was interesting because a lot of companies from here, from Europe, start writing to me. I remember the first time that I get contact to go to London to give a conference about data [inaudible 00:21:46], that companies like Square Enix or King.com or all those big companies, just look to my blog and what I was posting because nobody did that before.
Juan Gabriel: 00:22:01
And I went to those conferences and every time I went to one, I came to back to Mallorca with a new offer to go to one of those big companies and my boss only had one option, equal my salary here in Mallorca because otherwise I would have gone to London or to Canada or Tokyo, whatever, to work for those big companies. So imagine myself, with just 26 years old in that days, growing one step after the other and becoming the head of the product of the those video games at the company. I saw the future in that case. I saw that people really like what I was doing, really liked about this analytics and no one else was sharing the results like I was doing on those years.
Juan Gabriel: 00:22:51
So this is where I started thinking, “Okay, if I am able to do that in this little company, and those big companies are willing to spend $100,000 on me to go to this different places, and do the same for them, why am I not teaching that to the rest of the world? I could be able to create more positions for people to apply to and really democratize the knowledge of the machine learning and the data science.” And that’s what I did, I went to my boss. “Hey, I’m quitting, I don’t want to work with you anymore.” He told me, “Okay, whose company had now reached you to go to work with them?” And I told him, “No other company, I just want to be my own boss, I want to teach to the rest of the people in the world what I learned. I want to keep learning, I want to become a teacher.” And that’s when I quit, I started my first online course. I started the same way that I did, I started iOS development because I first created iOS applications, then some Android applications.
Juan Gabriel: 00:23:58
So the first stage was people need to learn to code. The second step was [inaudible 00:24:05] video games. I specialized in video games so I started teaching video games and that when I created my first Unity courses. And the last stage was okay, now let’s dig into data, let’s create some machine learning course, some data science course, even artificial intelligence course, and that’s been the last stage. In fact, when I met you, back in the Udemy Live, and we started our partnership of creative courses and translating courses so you see all the evolution from the beginning, I just wanted to be a doctor, to right now, where I am teaching what I’ve learned and I keep learning at every stage of what I do.
Kirill: 00:24:42
That is very successful you’re teaching, you have over 200,000 student, congratulations, it was very exciting to see. Very cool, and just for our listeners, you teach in Spanish, is that correct?
Juan Gabriel: 00:24:58
Yes, I never launched a course in English, despite everybody tells me I should do, I know that it’s also a very competitive market and I don’t see the point. There are already a lot of good instructors in the different aspects in English, in fact, 90% of the teachers teach in English so it would be really difficult to start in that market. So all my courses, I think 77 already are in Spanish.
Kirill: 00:25:29
Okay, and we’ll get to that in a second, about data science in Spanish. But first, I wanted to talk about your experience in the video game industry. I know you have a case study for us, from the video game industry. Can you share it with us, we’d be very excited to learn how data science was applied.
Juan Gabriel: 00:25:57
Yeah. The problem that we have in video games, as I told you, is that, when you have a free video game, you can have a lot of people playing that game, but it’s really difficult to get the conversion, to get one play to become a payer in that case. But you need a lot of payers to be able to have some money, escalate, get more people, to hire more people, and keep growing. So the biggest problem we have in the video game industry is, how do we convert a player into a payer? There is a lot of methods. We are to first engage the player once that is engaged, how to keep coming to the game and he just gets more into the gameplay and finally there is some point where we try to force a conversion in that case. But the main problem is the low percent stages. Do you know what is the percentage conversion rate from player to payer, usually?
Kirill: 00:27:06
No, I would guess like 1%, maybe?
Juan Gabriel: 00:27:09
It’s between 1% and 2%, yeah, you nailed it.
Kirill: 00:27:12
Wow.
Juan Gabriel: 00:27:13
That means that when you have one million players, only 1% will become a payer. So you have to keep an infrastructure that is able to get all these people playing with [inaudible 00:27:27] or whatever you have behind that game, to hold all that data and all those players with just such amount of money that it’s really difficult to escalate once you get there. So what did I do in this case? It was a slot machine video game. I decided to look at those players we already had. And in that case, not all the players behave the same way. For instance, some of them are more willing to share because they are social players, they want to share with their friends on Facebook, on Twitter, on wherever. So they’re really potential viralizers, they will probably not pay for anything on your game, but they are your best marketing team because you can give from free experience to those guys if they share to their social accounts. But this is really good to have more people without having to invest more money on your video game. So this is one kind of behavior we like to detect on the video game.
Juan Gabriel: 00:28:39
The second one is those that are really engaged to your game, they are super engaged, they’ve been playing for months, but not paying for anything. So you need to detect those kinds of behaviors in order to treat that conversion, they’ve been using a lot [inaudible 00:28:59] so what is the best way of dealing with that? It depends. Depends on what are their intrinsic needs, what is the shift they need to do to become a payer? That depends, some of them will be high rollers, they are willing to pay with a lot of virtual money so they need to do a really high bid to get excited in order to another one and another one and another one and another one until they run out of coins and they need to pay. This would be like someone that is really engaged in a casino, for instance, that is paying a lot of money to keep playing in that case.
Juan Gabriel: 00:29:38
And some others will be different, some of them will need little personalization, for instance, being able to write in the chat but change the color of the characters, that is really interesting for people that like to write or like to share content. And finally, there is one special case of people that is willing to look for a lot of things to get money for free and those are the special cases for advertisers. The advertisements are really cool because you can add one extra button, hey, look at that advertising and get one game for free. And this is really cool because when you try to put everything together, when you try to segment all your players into the different behaviors, and each behavior is given what it expects, you multiply your conversion a lot.
Juan Gabriel: 00:30:41
You don’t expect to have money just by the old-fashioned way, this 1%, 2%, but now you start having money for different places, from the advertisements, from these free people that is inviting more people so you grow your database a lot. And from the high rollers that want to invest finally in your game because you are giving them what they expect. So this is super interesting because it’s a problem of machine learning, you have to do some big clustering, after that, you need to decide what event is really useful for that and what is not. And after that, you have to create some automatization and neural networks that are able to fit the data, almost daily, in order to see exactly what is the best approach for each group.
Kirill: 00:31:39
Okay, very interesting, so did you already know these four different groups and then you applied machine learning or did you use clustering to discover these groups?
Juan Gabriel: 00:31:52
That’s the trick. Now I’m talking from my perspective so I know that this group exists. But back on the analysis, I have no clue how many groups there would be. And even if I would be able to give them a name or a label, that’s why we did a clustering, otherwise probably a classification tree or random forest or something, would have worked if I already had those labels. But back in 2014 or 2015, we didn’t have a clue how many group there would be. In fact, when I used this technique, we saw the different video games, I faced that there is not four groups, there are five or six or seven, that depends a lot on the general public that is playing to the games, the demography is really important.
Juan Gabriel: 00:32:41
I see that, for instance, Latin American people tends on be in this group people that like to watch an advertiser in order to get some virtual money on [inaudible 00:32:57]. And United States or even England is willing to go to the last stages when the high rollers live. And that means this 1%, 2% grows to 2%, 3%, not more than that. But you are doubling the money that you are earning and you are dividing by half the odds of losing a payer. So imagine that this 1% generates the 80%, 90% for the whole company, it’s like having a poker hand, if you lose on of the hands, you are sc***ed in that case. So the biggest it is this percentage, and the more able you are to average where the money comes from, the better odds in order to the company for success. And I’ve seen people that invest a lot of money in video games. There’s one guy paying $20,000 a month because he’s super engaged to a video game. That’s insane. But that is your poker hand, you lose that guy, you lose $20,000 a month, it’s a lot.
Kirill: 00:34:08
Very interesting. So effectively by making some of these adjustments with clustering and machine learning techniques, you’re able to double the amount of money that is spent and reduce the probability of losing a customer by two, that’s huge, that’s a huge result.
Juan Gabriel: 00:34:29
Yes, and even if we don’t talk about money and we talk about the engagement, you would be also surprised of how many people just download and just do one game and leaves the game. What’s the percentage do you think?
Kirill: 00:34:48
Oh, probably quite high, maybe like 50%?
Juan Gabriel: 00:34:54
Yeah, 50%, 60% of people, even up to 80% just open the game, does one game and then leave. So you can see that the potential group of people to monetize goes down with just a few minutes of game play. What does it mean? It means that is a long tail distribution, and the bigger it is the tail, the more chances to monetize them. So what would you do as a data science in order to increase that tail?
Kirill: 00:35:32
Well, I would somehow focus on increasing engagement.
Juan Gabriel: 00:35:36
The bigger the engagement is at the first stage, the slower the tail will go after some amount of time. So the critical stage is the first game, even if there is a second game, how do we do that? Tweaking the randomness. Tweaking the randomness means that the first game that do on a video game, it’s probably fake, it’s already prepared for you to win, and it’s already prepared for you to have a nice experience. The better your first experience is, the bigger it is the tail in the end. So I am devolving a lot of secrets from the video games to you and your audience, but you can see that a lot of the games always tend to let you win the first stage and you are not really playing with anybody.
Kirill: 00:36:29
Yeah, that’s interesting. It’s similar to online education. The first experience, the first few lectures really matter. If you, as an instructor, record very boring and not interesting first few lectures, even if later on you have the best materials ever, people are never going to get to it so it’s super important that at the very start, you show your students what they will learn and get them excited about it.
Juan Gabriel: 00:36:59
It’s the same principle. This is the gamification we were talking about, and this gamification is super important because the more engaged people are in one product, in one service, could be a video game, could be an online course, could be your podcast, the more chances it is that these people will come back and they will last longer on your product.
Kirill: 00:37:23
How can data scientists use engagement or gamification in their day to day job? So how can data scientist use gamification in their day to day job which is unrelated to the industry of gaming?
Juan Gabriel: 00:37:40
This is a really interesting question because for me, gamification is the glue for a success. The more engaged people are in doing on thing, the better odds of that becoming a success in the future. So in that case, when a data scientist is trying to do some kind analysis or whatever, he can think in terms of [inaudible 00:38:07] about a simple ranking. For instance, a ranking is a way of gamifying things, and try to do three, four, fives, different algorithms, establish some score, could be r-squared, could be p-value, could be whatever you want, ratio success. And when you run them out, you say, “Okay, this is my ranking, my best algorithm is this one and the worst is that other one.” Let’s pick the worst one and let see if we do some fine tuning is able to escalate some steps far beyond the first one. And you do it iteratively and you try to investigate more algorithms and more ways of doing things, until you have this super ranking that is not able to beat anybody at the other.
Juan Gabriel: 00:38:52
It’s more or less this kind of competitions, they are gamified because you see this ranking technique that they apply, you say, “Okay, who’s the best one, solving this problem?” That’s the same batch for your own problem. Also, it’s very interesting because if you work with group of people instead of being yourself, you could use some achievement system, even some badges, you could start gamifying like a video game would be, and that is super interesting because you tried to always beat the other one, or get this achievement, or get this badge that you don’t have, by thinking out of the box. The problem I see in data science is that, when you are able to do three, four, five algorithms, we try to solve everything with those algorithms without investigating farther out of there. This is like knowing how to use a hammer and driller from your toolbox and try to use that for everything, so when you have to take out the screw, you use the hammer or you use the driller or something like that, instead of going to the screwdriver. So it’s like that, but in terms of machine learning or data science. Knowing five or six algorithms is good, be able to investigate or take a step further, do some fine tuning, reading some papers, is going the extra mile, and this is gamification.
Kirill: 00:40:20
That’s gamification for myself, right? Learning new things for myself. How about gamification for the clients or for instance, stakeholders. If you’re doing a presentation, often data science presentations or results reports can be kind of dry. Is there a way data scientists can use the gamification to engage their audience, the stakeholders of the project, more?
Juan Gabriel: 00:40:47
Well in fact, I did it with you. I asked you two times, what do you think in that stage of this and that and the other one. Let’s make them part of the presentation, ask some questions. Answering the question can be a rewarding system when you answer correctly and can be a really good way for you to teach something that they don’t know. The typical problem, when you deal with a client, the language of the client and the language of the data science is not the same, they are thinking some things and you are thinking anyway different. It’s that, tell me what you want and I will show you what you really want. It’s a different language. So doing that during a presentation, “Okay, what do you think is this ratio? What do you think is the best algorithm? Or between this three game profilers, which one do you think generates more money, the people that pay or the people that watch this advertisement?” This is a communication. You are not lecturing, you are teaching, so this is important because this communication helps a lot on the engagement or on the gamification of the situation. That’s just one example, there could be a lot more.
Kirill: 00:42:00
That’s very interesting. I realize now when you asked me, how many do you think percentage of people convert from players to payers, so that an example of gamification, indeed it makes me more engaged in the conversation.
Juan Gabriel: 00:42:15
You do it a lot. When you create your courses, you try to anticipate the student to what will happen. Okay, this is what we have right now, what do you think will happen, the client will purchase, will drop out, will become video watcher, advertiser watcher. You try to anticipate when you do your courses or I do mine. We do it so naturally that we are not really conscious of what’s happening behind that, but that’s what it is, we are creating a gamification experience in teaching as well.
Kirill: 00:42:50
Gotcha. And speaking of teaching, you’re not only teaching on Udemy, you also teach at the University, right, you’re an associate professor. Tell us a bit about that, how did you get into becoming, teaching at uni?
Juan Gabriel: 00:43:04
That’s interesting because at the same moment I quit my job, and I started teaching online, I got a call from a teacher at the University, “Hey, there is one option for you to come back to the University, there is some linear algebra classes that you give. It’s not a lot of time just four hours per week, if you’re interested of coming back to the University.” When I quit the master in Paris, I lose all the opportunities to come back to the University, I was able to that masters because this University here was able to get me this scholarship to go over there because of my note or my behaviors or whatever, but they held me on that. When I left this master, I break all kinds of relationships with this University.
Kirill: 00:43:55
Yeah, they must have been quite upset.
Juan Gabriel: 00:43:58
Yeah, they were because I was supposed to come back from this master, teach them what I’ve learned in Paris and all that stuff, that’s didn’t break out when I come back.
Kirill: 00:44:09
For a girl.
Juan Gabriel: 00:44:09
So they call me, “Hey, there is this opportunity.” Okay, let’s try that, four hours per week it’s not a lot. Let’s see how it goes. And it was really related with my online teaching that I was starting back in 2015 and I think it’s the best I’ve done, because when you teach online, it’s like doing this podcast and not seeing you, and not seeing your face, not seeing your reaction, and it’s really cold, you don’t know what the other one is doing. Could be sleeping, could be looking at their mobile phone, I don’t know what they are doing when they take one of my online courses. But when I teach at the University and I see all the faces, I am able to see, okay, now am I boring all the audience, so let’s make a change or this is not the correct way of doing things, this is too complicated, lets make it simple. And this is the best feedback I’ve had for the last five years because by watching the different faces, year after years, despite I am teaching the same subject, I am able to improve it, to tweak it, and even get best results year after year.
Kirill: 00:45:22
Wow, that’s really cool. And I think anybody can replicate that in their career, even when you’re presenting to people, whether it’s live in the real world, or virtually, now with the situation we’re in. Try to observe that feedback or even record, ask someone to record the presentation, of course ask permission from the people in the audience, but if possible, record it and then watch it later and see how you sound, how people react and all these things. It’s very powerful, as you say, to analyze yourself. Let’s just present and okay, it’s done. But actually go through the painful experience of listening and watching yourself and understanding where you can do better.
Juan Gabriel: 00:46:12
Yeah, it gets you a real nice way to watch of what you do, what are your gestures, how you move your hands, how you move yourself, how do you position yourself. If you are becoming lazy and you start to getting one side or the other of wherever, it helps you improve your position talking in public. And it also helps you to become a better teacher. I always say that I don’t teach for the smartest one, I teach on the other side for the dumbest one. If the one that has the more difficults to learn, is able to learn, all the others will also be able to learn. If you teach for the smartest one, only this guy and smarter on the room will follow you, all the other ones are lost.
Juan Gabriel: 00:47:01
It’s like this gamification experience we were talking in video games, this long tail is aim to improve the first experience in order to improve the tail. The same thing happens when you teach. The first moment you go into the class, the first topic you teach, are the ones that will decide if you engage in this audience or they will drop out. So the final goal for them to pass your subject or for you to teach [inaudible 00:47:29] linear algebra, calculus, machine learning, data science, whatever, but it’s these first moments and these first experiences that you need to be aware of, because otherwise, you will lose them during all the class.
Kirill: 00:47:42
Okay, that’s very interesting. Speaking of students and learning, what’s the difference for someone learning data science at a University versus online? What are your comments on the two, because you’re in a unique position that you teach in both cases. So do you think online education will take over University education with time or is there a place for both?
Juan Gabriel: 00:48:11
Right now in fact, due to the coronavirus things, I see it multi-layered before. What is the purpose of having a University right now? To expand on paper, on paper that say Juan Gabriel has a mathematics degree or has this teaching education masters, whatever. Is the whole purpose of having the University this paper? Because the best teachers are not on a University, are outside the world with their experiences and they can share their knowledge through online courses for instance. So what I’m facing right now, I will use my same online courses in the teaching of the University, for instance, I teach at the master of data science in Mallorca, and I use our machine learning from A to Z in Spanish. That’s the topics we cover for my subject. And they’re super excited to have all the content online, all the community, access to the forums with the questions of the other people, that I don’t see the point of going to the University, to the class, and teach that with a board or projector or whatever.
Juan Gabriel: 00:49:25
So what I think is that the University will be the only way we will have to expense from official papers that says you have this masters or this other thing. But, as soon as more companies realize that this paper, it’s not worth, and what is worth is the abilities, the things that you are able to show on an interview, the things that you are able to do when you work on that company, they will have to make a shift because online courses will be as valuable as a title from a University. That’s what I think.
Kirill: 00:50:03
Mm-hmm (affirmative).
Juan Gabriel: 00:50:04
I don’t know if you agree with me or not?
Kirill: 00:50:09
I like your thinking. I think the Universities have a lot of reputation behind them, and a lot of brand recognition and they can use that in a hybrid model. For instance, a University could partner up with a company that already does online education and the University could look through the education and if they’re happy with everything, they could put their stamp behind it, and create these hybrid models where, I think the sooner University start realizing this and getting into it, those Universities that start first will really expand because why not? Why not learn online and get the certificate of approval from a Harvard or a University of Mallorca or somewhere else. That’s two birds with one stone.
Juan Gabriel: 00:51:06
Yes. The other problem is that Universities are really conservative and they want to have their own teachers, they want to have their own materials, their own brand, as you say. And I think it will take a lot of time in that case. I think that this coronavirus will make a change in the education because right now, probably 2020, 2021, 2022 will have to adapt to an online education and there will be two ways of doing that, you have to create your own materials from scratch despite you don’t have the confidence to talk on camera or you don’t have access to this kind of way of teaching or you are not able to do that. Or go to some materials that are already created by someone around the earth, and use those materials. You could have access to the best recording of Kirill or the best recordings from teacher of Harvard or something like that, with just one click away. And I think that’s the way to go. As you say, if you already have these materials, if it was a book, you wouldn’t write again your book, you would recommend that book that has already the materials. I think that with online courses or videos already recorded would be like that in the future, or at least I hope that it is like that, which is the natural way of expanding of there is already good materials, why to reinvent the wheel twice.
Kirill: 00:52:33
Yeah, absolutely, agree. Just talk a bit about the data science world in the English speaking world versus in the Spanish speaking world. What have you observed, you’ve been part of both. Again, very unique position. How is data science developing? Is it different in the Spanish speaking world or is it exactly the same as in the English speaking world?
Juan Gabriel: 00:52:57
I think there is not a lot of difference because a lot of algorithms start with papers written in English and they adapt even with English terminology. In Spanish, NLP, natural language processing would be Processamento de linguages naturale, which would be PNL, the other way around. But we don’t use PNL, we use NLP, it is the natural way of talking on how you learn all that stuff. This is just one word.
Kirill: 00:53:24
One that really confuses me is [Spanish 00:53:27], right? Is IA. And I see a lot of people using IA.
Juan Gabriel: 00:53:35
Yeah, well there are a lot of people that translate these small word but I don’t see the point of doing that. If you just say NLP in Spanish about artificial intelligence, we’ll be able to follow your speech. So there’s not really a lot difference. The problem that I see is there is a lot of people that don’t know English, their lack of basis in English is what they are facing right now, and they’re their biggest mistake perhaps that they should know more about English or how to read the paper or how to read the documentation, Python documentation is in English, R documentation is in English. And sometimes I see a lot of people just copy-paste on Google to translate to Spanish instead of, okay let’s want to take the extra effort and learn a little bit about English so we’re able to follow the whole meaning at least of the function you are using or the paper you are reading. So that is the only difference I see.
Kirill: 00:54:38
Okay. And how about in terms of students? So you have 200,000 students on Udemy, do you see a growth in the number of interest in the space of data science. For instance, in the US, it exploded, huge rise in job offers, job demands and there’s a lot of people wanting to get into data science. Has that been the same in Latin America or in Spain or is there a lag behind it, are you seeing the same explosive growth? What’s going on there?
Juan Gabriel: 00:55:14
I think it’s two, three years behind so this explosion has just begun one year ago, more or less. But the point here is that data science is a general world for so many different things that in Spanish, it’s even wider, because it could be a marketing director doing some Excels, and he could call himself data scientist, in that case. So it’s so general a list at the first stages that talking about data science is just talking about using Excel, in that case, or using technology. So in this case, I think that, yeah, we are really behind the English market, in that case. But I also see that month after month, a lot of people is asking about this, what’s the difference between machine learning, artificial intelligence, data science, even learning mathematics, because I always say that mathematics is really important to know what you’re doing. The biggest problems I see with people that is taking machine learning or data science courses without the basic. I don’t say that you will need a degree in mathematics, but it’s important to know the basics.
Juan Gabriel: 00:56:36
The biggest problem I see is that they don’t know about the elementary things that apply one’s technique to some piece of data and they have similar matrix problem. Okay, what’s happening, what I did wrong? The problem is you don’t think about the data you have, probably you have [inaudible 00:56:57] or any kind of problem behind that, that is you don’t know what that means. It’s just applying some recipe without knowing the ingredients you have. Is it? So that is the problem I see that a lot of people want to go into data science, want to go into machine learning, but they want to go the fast way around instead of having all the basics that then you will have. Even I face people that, “Hey, I don’t know how to code, may I take your artificial intelligence course?” The answer is no. For sure [inaudible 00:57:32]. Or else it’s like trying to learn Japanese without having the basics or trying to write a book in Japanese without knowing the language.
Kirill: 00:57:43
Okay, very cool. And so, you’re seeing that in the Spanish market that people rush into data science like that.
Juan Gabriel: 00:57:54
Yeah, and day after day, I see it more. Because I see a lot of people follow me on LinkedIn and a lot of people that says data science in blah blah blah, data science in that company and the other and the other. But when you take a look at that, you see that these people that never did an analysis or at least they don’t share anything about data or they don’t follow any website or they publish about data analytics. So that’s interesting because you call yourself data science or your position in the company is data science because someone at human resources told you so, but later on, you don’t do data science you just do an Excel compellation or you just grab data from a database and you put it on a file, that’s it end of story. So I see that this is a really misused word in the Spanish market at least.
Kirill: 00:58:47
I think it kind of consequence of that the boom is just happening now and a lot of people still don’t understand what it means and I think with time, even in the English market, it still requires time to become more formalized and clear. But there’s a lot of opportunities in that. When there’s not enough structure, that means there a lot of inefficiencies and that means you can really stand out and so following your advice, just doing it methodically, learning all the fundamentals, understand what you’re talking about and then going into it, you can really be the data scientist that everybody desires rather than just a wannabe data scientist, if that term can be used.
Juan Gabriel: 00:59:34
For sure.
Kirill: 00:59:36
Sounds good. Well, Juan Gabriel, I can’t believe how fast this hour flew by. It’s already been an hour. I wanted to ask you, I guess one more question, before we start wrapping up and that’s, given all your experience and all the things you’ve seen in your career, where would you say we’re going in terms of data science? What’s the future in the next three years in the industry? Not just in the Spanish market, overall for data science? And what should our listeners do in order to prepare for the future that coming?
Juan Gabriel: 01:00:15
I think that we’ve reached a point where all the technologies have been applied to the different fields on machine learning or artificial intelligence that now in 2020, [inaudible 01:00:29] toolbox we’re talking about, we have a super prepared toolbox to solve any kind of problem more or less. So I think that this is the perfect problem too, start learning how to use this toolbox in the different aspects. I see a lot of companies that, were worried about data, about data science, about machine learning, a few months ago. But after this coronavirus, they need to start worrying about this data and apply correctly these techniques. So I think that on the next years, it’s the perfect moment to start applying all this knowledge on machine learning or on artificial intelligence to real problems, to business problems, to industrial problems, to medicals problems like these virus around. And I think we probably won’t see an improvement or new techniques, not at least as big as we saw on the last few years, despite probably from Google, OpenAI or something like that, that they always invest in new R&D project. But for the usage of these techniques, I think it’s the perfect moment for start using all these toolbox that these last decade has provided us.
Kirill: 01:01:55
I love that. That’s really cool. It’s now focusing on the usage. It’s like the trade off between exploitation and exploration in AI, right? We’ve found a lot…
Juan Gabriel: 01:02:08
Yes, it’s exactly the same. Now it’s the moment to start using those things and I think that the coolest one that everybody should know about is the NLP and the Chatbot and automatization, I think that with the latest algorithms, now is the perfect moment to automatize things that would be done by a human. And I’m talking really big on that. Here in Spain, for instance, the taxes are still done manually. There is one person reviewing everything, okay, this is an expense, this is not. This can go this way, this can go the other one. Well, now is the moment we start relying on these difficult job and also human error job with automatization on the AI side and I think that creating Chatbots, creating automatic counseling, even for us, on online learning, having the possibility to answer automatically a question, by looking for all the previous question or even the transcription of our own video, would be interesting.
Juan Gabriel: 01:03:19
For instance, yourself on the podcast, what Juan Gabriel thinks about the future of data science? Asking on question like that, and one bot being able to prompt to the exact moment where we talk about those things. I think that is the perfect moment because we have the technology, we have the knowledge, now is the moment to use that for something useful. And I think that’s the perfect [inaudible 01:03:43] for somebody to start into the world of the artificial intelligence or the data science.
Kirill: 01:03:50
That’s absolutely right. Yeah, so very interesting, there’s lots of cool things, lots of scary things that can be done with NLP, right? I heard of a story where, this was just not in NLP, this is like a proper deep-fake with NLP and video when somebody in Canada called an accounting department of a company and it was the CEO calling, I don’t know if it was with video or just phone call, and they said to pay $200,000 to an account, and they moved the money and it was not the CEO calling, it was a fake image so, it’s good things and bad things that can be done right?
Juan Gabriel: 01:04:35
Yeah, last week I was reading also about, I think it was Jordan from your team, that sent me one link about how an artificial intelligence wrote a whole post and it was marked as the hack of the decade, they felt that it was a real news, but it was absolutely a fake news generated by some NLP using this GPT3 that is so powerful right now.
Kirill: 01:05:03
Fake news.
Juan Gabriel: 01:05:04
Yeah.
Kirill: 01:05:06 Wow, GPT3, yeah. Cool, okay. Let’s see how it goes right? Let’s see where this all takes us, it will be interesting.
Juan Gabriel: 01:05:16
Yeah, this is like the change of the century probably, back in the 19th century when they started discovering the shift from the steam machine, later on the electricity, all that stuff, relying on a lot of work and created new kinds of positions. That is what will happen. Why would you have a lot people reviewing the taxes one by one or why would you have a lot of people one by one, if we can automatize these things with artificial intelligence or machine learning, for sure, those people will have to learn new things, probably management of those techniques or fine tuning of the parameters behind those algorithms, they have to switch from one job to the other, it doesn’t mean that we are taking these people out without any kind of job, no, that means that they adjust to the role like all these algorithms are evolving during the last year.
Kirill: 01:06:18
Absolutely. It’s just a question of speed. Before electricity and steam engine, the speed was not as high, you could change a job and have time to learn. Now, this is just going to happen so fast, will people have time to learn and re-qualify for a new job?
Juan Gabriel: 01:06:40
Also, jobs lasted for 35, 40 years, 50 years working on the same position for a person is a lot of time. Right now, there is no person working more than three, four, five years, even in the same position because the position changes, the company or even the technology, the technology changes so fast that we have to be updated the more than we can.
Kirill: 01:07:04
Absolutely. Okay, Well, Juan Gabriel, thank you so much. Before we wrap up, where is the best places to find you for our listeners?
Juan Gabriel: 01:07:14
They can find me on LinkedIn, it’s Juan Gabriel Gomila Salas, it’s in Spanish. It’s easy to recognize me. And also on my website, frogames.es, where they can find everything we’ve talked about, they have my Ted Talk, they have all the online courses, even the learning paths for the different careers, like video games, like machine learning, data science, even mathematics. And for sure, if they went to follow my travels, I know that I don’t travel right now, but my possible future travels, I post a lot about the different place I go. Facebook, Twitter or Instagram is the perfect places to find me. Again, look for Juan Gabriel Gomila and it’s easy to find.
Kirill: 01:07:56
Awesome, fantastic. And Juan, last question, what’s a book, I know you wanted to recommend several books, so what are the books you like to recommend?
Juan Gabriel: 01:08:06
Yeah, they go really related with the story I told you about how I started on all those stuff. And there is one for each kind of topic we covered. The first one is called, Heads First Data Analysis, it’s a really old book from 2009 from Michael Milton. It’s super cool because it shows you about the basic knowledge you need to know to start on data science. You start doing a simple linear regression then the story goes on but this regression has an outlier so the next month, the model is not accurate and you need to learn how to identify outliers. When you have that, you start learning new things. And it’s super cool because it’s a whole story from a guy that is learning data science from their beginning doing all simple linear regression, to the end, presenting the final job to the boss and the successful data science. It’s super easy to read and this is one that changed my perspective from mathematician to data scientist.
Juan Gabriel: 01:09:13
The second one, I think that you recommended that also, is called, The Lean Startup, it helps me a lot when I started creating my own course and becaming my own boss because it helps you to identify this possible needs of your business to how to talk to the different clients, how to be aware of everything new and how to manage your own company, so I think it’s a book that everybody that’s an entrepreneur, should read about. It really gives you a little clue about what’s happening and how to face all these different things that you have to deal with the entrepreneurship.
Juan Gabriel: 01:09:56
And the last one, it’s called, Gamification By Design, from Gabe Zichermann, which is super interesting because it’s the things talked about, gamification applied to everything. It talks about game theory, which the field of politics and economics used a lot. And how that game theory is applied to the different aspect up to problem of the real life, like gamifying the education, gamifying the management of your team, that everything that happens on your company. So I think it really a nice reading for all the people who want to gamify their own experiences day by day.
Kirill: 01:10:39
Wow, thank you. So we got Head First Data Analysis, The Lean Startup, and Gamification By Design.
Juan Gabriel: 01:10:47
That’s it.
Kirill: 01:10:48
Awesome.
Juan Gabriel: 01:10:48
All available on Amazon and that stuff so they’re really easy to find.
Kirill: 01:10:54
Cool. Thanks for your recommendations Juan Gabriel. And yeah, on that note, thanks a lot for coming on the show, it was really fun chatting to you.
Juan Gabriel: 01:11:03
Yeah, thank you very much and I hope that we will keep collaborating because I think that I learnt a lot from you, I haven’t said that on the show, but I learnt a lot from your courses and this collaboration, porting these algorithms to Spanish has been a way for me to keep the track of machine learning or artificial intelligence and even gamify my own experience. Over there, on my studio has a blackboard with Kirill’s course completed, 25%, 28%, 29%, this is the completion of both the course and the translation. And this is my gamified experience to keep translating courses into Spanish and learning as well.
Kirill: 01:11:48
Fantastic. That’s really cool to hear, you’re very organized. And I also hope we will continue, thank you for translating so many courses, that gave access to them for so many people. Just today, I saw that your Spanish version of Machine Learning A to Z has I think 11,000 reviews, just the reviews, it’s huge number of students there. And it’s also very great to hear that or see that more and more people can access this knowledge.
Juan Gabriel: 01:12:18
Yeah. I see a lot of students that thank me for translating the courses because it’s like one of the sentences, it’s like democratizing the knowledge behind the machine learning or the artificial intelligence and being able to approach that to people that don’t know English or are not able to follow a class in English. And I think that is the future, or the key of my success, at least it has been, to port materials or create materials in my own language, in Spanish. In 2015, there was no course in Spanish on Udemy, for instance. So I think bring the knowledge to your mother language so you are able to understand everything clearly and become data scientist or an engineer on video games or on artificial intelligence.
Kirill: 01:13:10
Absolutely. All right, thanks you very much. I look forward to seeing you in Mallorca, yeah? Hopefully after all this is gone.
Juan Gabriel: 01:13:18
Hey Kirill, we have a barbecue here when everything goes normal so you’re invited.
Kirill: 01:13:24
Awesome, I’ll see you there.
Juan Gabriel: 01:13:27
Okay.
Kirill: 01:13:31
So there you have it ladies and gentlemen, that was Juan Gabriel Gomila, hope you enjoyed this conversation, you got energized from it. My personal favorite takeaway was the whole notion of gamification and applying it, not just in the gaming industry, but overall in your life, in your career and it was really cool how Juan Gabriel then illustrated, it was like subtly applying it in our conversation and then illustrating how he was indeed doing that by asking me questions like getting me to guess those percentages, how many people convert from player to payer and that other second time. That really helped kept me engaged and that’s a way you can also help people be more engaged in your presentation. Think about it next time you’re presenting something to your team, to your manager at work, see what kind of questions you can ask your audience to get them engaged, I think that’s a very powerful tip.
Kirill: 01:14:34
And as always, you can find the show notes as www.superdatascience.com/403. That’s www.superdatascience.com/403. Juan Gabriel teaches a lot of courses in Spanish, data science courses in Spanish so we’ll put up the top three of his Spanish courses in the show notes, that link. So if you are learning data science in Spanish, then check it out. Maybe that’s your mother tongue, he has, as I say, some of our courses, he’s got courses of his own and he’s a fantastic instructor for over 200,000 students on Udemy, so that serves as a testament to his way of teaching, make sure to check them out. And also if you know somebody, if you just know someone who wants to learn data science or is interested in the space but maybe language has been a barrier for them, maybe their mother tongue is Spanish and they haven’t been able to find quality courses on the topic, well this is your chance to help them out. Just send them the link, www.superdatascience.com/403, and once again, we’ll include the top three course by Juan Gabriel there and they’ll be able to proceed to those course and check them out for themselves.
Kirill: 01:15:52
There you go, share the love, send that to one of your friends. And as always, thank you so much for being here today, super pumped that you’re continuing with the SuperDataScience podcast and I look forward to seeing you next time. Until then, happy analyzing.
