• Ken Jee

Can Data Science Be Beautiful? (Thu Vu) - KNN Ep. 105

Updated: Jul 18


Today, I had the pleasure of interviewing Thu Vu. Thu is a Data Science Consultant with 5+ years of experience in data analytics + data science. She has a background in Economics and Computer Science, focused on building tools to communicate data insights. She also creates videos on her YouTube channel to share advice and help newcomers in Data Science develop their skills. In this episode, we learn about introspection and how it can help you find more enjoyment in your career. We also learn about how art and beauty have influenced Thu's career and content.

 

Transcription:

[00:00:00] Thu: Another side of data science is also doing things that no one is actually doing or like solving problems that no one had actually thought about using data to solve.

[00:00:20] Ken: This episode of Ken's Nearest Neighbors is powered by Z by HP. HP's high compute, workstation-grade line of products and solutions. Today, I had the pleasure of interviewing Thu Vu. Thu is a data science consultant with 5+ years of experience in data analytics and data science. She has a background in economics and computer science, where she focused on building tools to communicate data insights. She also creates videos on her YouTube channel to share advice and help newcomers in data science develop their skills. In this episode, we learn about introspection and how it can help you to find more enjoyment in your career. We also learn about how art and beauty has influenced through's career and content.

I hope you enjoy this episode, I know I enjoy talking with her. Thu, thank you so much for coming on the Ken's Nearest Neighbors Podcast. Very excited to have you here. I've seen your YouTube videos. I've talked with you a couple of times before this, and we were actually introduced through our mutual friend, Luke Barousse. So again, excited to have you on here to talk about, you know, sort of a different perspective, obviously you are over in Europe doing data science there, and I'm really excited again, to hear your story and how you got interested in data, as well as YouTube.

[00:01:29] Thu: Yeah, thanks, Ken, for having me today. Yeah, I'm super excited as well. Yeah, thanks to my friend, our friend Luke that I got to know you, and, of course, I've been following your channels for a while now. And yeah, like to, yeah, when I start in getting into data science, it must have been 6 years now, already. And I actually, it was a little bit unintentional back then, of course, in Europe in the Netherlands in 2016 everyone was talking about becoming a data analyst.

And that was probably the hardest job back then before we started having the hype about data scientists and machine learning engineer. And so 2016 was when I graduated from my master program. So I studied economics and also in my bachelor, I studied business and economics. And so. But I already knew from inside that.

Okay, I'm relatively good at programming and I like tinkering with software and with coding. So I knew that yeah, data analysis and data analysts sounds quite interesting to me in terms of like, you get to work with a large set of data and you get to discover some hopefully very valuable insights and you also get to do some coding and so do some data visualization.

And I just love doing, yeah, I just love creating pretty stuff, so that is a kind of like a good combination of a lot of things that I'm interested in. And so just naturally, yeah. I was drawn to drawn to this kind of job. And after I graduated, I also yeah, did some internship, which is very much into like economics modeling and yeah, basically just doing some economics model not super advanced anything, not machine learning, more like regression or logistic regression and things like that.

So yeah, gradually, yeah, it's just build up some skillset. And so in 2016, I started my full-time job. And become, yeah, just be a data analyst, a junior data analyst, and to be precise because I had probably zero experience in data visualization and that sort of things. And so my R programming skill was very, very basic. Yeah. But it was a very interesting turn to me. So yeah, that's how I got into this field and be in this field until now and still kept me entertained kept me interested in learning new stuff. And, yeah. So I guess it's a good thing. Yeah, that I got into this field.

[00:04:39] Ken: Well, so what is it that you like most about the field? Is it there there's constantly new things to learn or is there something else associated with that?

[00:04:48] Thu: Yeah, I think it is both. I'm of a person who is very much. Yeah. It's, it's very easy for me to get bored after doing something for a while. So I guess, yeah, data science is just so fast moving that I thought. Wow. Yeah. I always feel like imposter and that's probably a good thing because I kept keep me, myself learning new things and it's also, yeah. Just like creating something valuable from yeah. Like just a set of data. I think it's very interesting. And I guess it probably resonates with a lot of people because yeah, we are, for the most part, quite curious human beings and just discovering something from kind of like.

A massive kind of like abstract things like data is yeah. It just makes ourselves feel like, yeah, we are doing something well good. Something valuable there to the world. Yeah. I guess, yeah, data science is just a very like among, it's just one of the very among, yeah, a lot of different fields that could contribute something good to the world. Right. And why specifically this field, rather than being, becoming a doctor like a researcher, or pilot or whoever that is I don't have a, like a good a good idea about why. Yeah, I'm more drawn to this, but I guess, yeah, more the curiosity and a little bit creativity, if not a lot. That comes to yeah, working with data and visualizing data and how to tell a meaningful story and make sense of stuff that we see in the world. Yeah, that, that is very interesting to me.

[00:07:07] Ken: I find that really, really fascinating. So, you know, some things you didn't describe when you were talking about why you enjoyed it was the technical programming aspects, the pure machine learning element. There was a lot about creativity, visualization, creating value. Do you think that there's too much focus on the more technical elements and there isn't enough? Conversation around the creative elements or the subject area related elements of the domain.

[00:07:38] Thu: Yeah, absolutely agreed on that. I also think that there have been, yeah, too much focus on technical learning, technical skills. I also see in my YouTube channel, a lot of people asked about, Okay, how should I learn? How should I learn Python or how should I learn SQL and things like that. And of course it's is where you probably, where is probably the best start like the best place to start, right.

To learn some like some, yeah, tangible skills that you could yeah, you could do something with it. Programming something or make a model or create a data model or whatever with yeah, SQL in the backend things. But I think, yeah. I don't think that it is the core of doing data science and from my experience also.

Yeah, I was, yeah. At my job, I was, yeah, always, I had always some feedback from colleagues that like, are you hiding too much behind your technical skills, but you really need to like, think about the business problem and to ask the question why we are doing this thing and how to approach it and to understand, Okay. What the business actually actually wants and want. Yeah. What kind of questions is the good questions to ask? Those kind of things. I also previously I also neglected it. I, so it's like, Okay, someone ask me a question and I just give them what they want, do analysis or make a machine learning model or whatever it is.

But I didn't think about, is it a good question to a good problem to solve? And is it really the problem or is it something else? And so, yeah, the domain knowledge and the things that are really, really important, I think. Yeah, it is a little bit like often on the blind spot of yeah. Data. Yeah. Doing data science because it's very, very yeah. It's not easy to, it's not easy to quantify. It is very, yeah. Domain specific. And it is very people also people specific. Yeah, so. I think like technical skills are good, but it's not everything. And more and more focus should be on the, on the soft side of the problem that is to to be creative and to be, yeah. Think out of the box. If the question is actually a good question to answer.

[00:10:30] Ken: So is the question a really good question to answer? I think that that's something. Is really hard to teach. Right. It's really hard to teach someone to ask good questions. It's really hard to teach someone to think critically.

It's really hard to teach someone in some sense to. A visualization, more beautiful or more compelling or something along those lines. And that goes all the way back to when they're learning it right. Technical skills. Some of them are, are, I mean, they're technical. Some of them are, are difficult to learn, but it's a very straightforward path to learn technical skills.

You have resources that say, Hey, you learn this, you learn this, you learn this. And it's very well laid out to learn the softer side of skills. It's a lot more difficult to quantify. And I think that that makes it intimidating or it makes it harder to measure. And that's why people focus significantly less on developing those because the way that we measure them, the way we evaluate them is so much more crude in some sense.

Again, that's not to say that they're less important, but it just means that they're harder to measure, which I think for some people is really terrifying, but we also have to remember that in the real world, when we're talking to people, we don't quantify everything. When we're in an interview process, when we're solving one of these problems, not every step is quantified and unfortunately, or fortunately, even very large decisions are not completely quantifiable.

If a business wants to, you know, release a new product or something along those lines. Yes, there are quantitative metrics that we're gonna use. But at the end of the day in a lot of companies, a human makes that decision still. They're not, it's not just like, Oh, we're going completely off the numbers.

It's like, Well, there are these things that the numbers can't completely, it completely evaluate. And so we need to, we need to build in our ex experience and our expertise over time. So I completely agree with you. I really would like to. Understand. How people can build these skills out in a more quantifiable way, or if we could bucket them into different things.

I think I made a video a while ago related to that, but apparently it wasn't that good. A video to be able to add more, add more insight into that. Let's let's change gears a little bit and talk a little bit more about value and creating that with your work. So we've, we've talked offline quite a bit, and you have a, a real focus on trying to do good with data. Can you explain to me a little bit more about what that means?

[00:13:08] Thu: Yeah. So it is a very, very good transition to this to this topic because I think, yeah, part of understanding the domain and understanding the right question, I think, yeah, it would help us a lot with like knowing how to solve real world problems that are like meaningful and in a way.

A lot of us, yeah, being a data scientist or data analyst, we are contributing somehow to, yeah, to the business. Right. We are creating insights like about the sales, about yeah, like the default of the customers for our company. However, yeah, I also think like there's probably too much probably like there's too much focus on, on that side of data science, but another side of data science is also doing things that no one is actually doing or like solving problems that no one had actually thought about using data to solve.

And I think, yeah, that is something, I guess, yeah, people who know about data science and also have a little bit of knowledge, knowledge on a specific field. I think we could try a little bit harder to reach those fields that could do so much better, could solve so many problems.

By just using a little bit of data insight, like two years ago or three years ago, I did a hackathon with some of my colleagues. So basically we met up with a group of doctors from a university here in the Netherlands and we tried to to help them to dive into that huge gene sequencing data of their clients.

So they are like oral lung cancer clients patients. And they, yeah, the doctors were trying to basically to find a way to look at the data and to be able to signify if patients is having yeah the more yeah, the bad type of cancer or a good type, like the malicious one or the benign type of cancer.

And so, yeah, we tried to do it in like two nights or two days and two nights and we didn't really sleep and it was really fun. And in the end, we got some really, really interesting stuff with it. And we didn't actually do any machine learning, but more like ed and really like rigorous rigorously looking at a lot of different genes, a lot of yeah, from a different angles.

So that, yeah, in the end we got some really interesting stuff coming out and it became a paper no academic paper after that. So I think, yeah, we could probably get out of our way a little bit more to solve these problems that we have absolutely no idea. But we could use the domain knowledge of some people and collaborate with them and to, yeah.

To be able to work together, to solve that problem. And it doesn't need to be like a fancy machine learning model or deep learning neuro network model. It could be very simple and in our reach. Yeah. And so I think that is probably a case for me to say, Oh, we haven't really done a lot of good things or I would have lied to do more of that myself to, yeah, just to contribute a little bit more to the fields that are, yeah, not very, very mature yet in using data science to solve the problems.

[00:17:09] Ken: That's awesome. You know, it's interesting, I haven't seen that many ways to be able to do that outside of Kaggle. I recently put together, I was looking for additional resources on where we could find some of this data and some of it's available through the FAO, which is the, I can't remember it's by the UN they have a lot of agricultural data, a bunch of these different things. And a lot of it is government release data, but a lot of it isn't very cleanly organized.

I'm surprised that there aren't a lot of non-profits that I've found that released their data publicly, which I think would be a really valuable thing for them to produce. Do you think the best way to find these things is maybe through Kaggle, through hackathons, through something else? Or is there a channel that I'm forgetting.

[00:17:56] Thu: Yeah. I think like looking at it online, I think there are so much, so much data and so many free datasets that you can find online. However, I think yeah, it's a little bit difficult because even if you get to work, yeah, you found a very nice dataset or like interesting dataset, but if you don't know that field very well, then it's also kind of like hinders you from actually be able to like, to exercise your, your science knowledge, because you don't know what to look at, what is important.

Even like what kind of questions you should you should ask on this dataset. So I think like, yeah, it's best to talk to people. Yeah. Hackathons would be very, very, a nice yeah, a nice way to get, to meet a lot of people from different backgrounds. However, I think, yeah, if it's too difficult or if it's like, yeah.

During COVID, for example, we couldn't really like participate in those kind of access activities then I think, yeah, just maybe like contact the local organizations near your place in Amsterdam here and also, yeah, in some other cities, there are a lot of organizations like agriculture organization or like auction house and all that kind of places.

And yeah, I haven't really done that myself, but I thought, Well, if I were to make a portfolio project for myself. I would probably go like a little bit like an off the beaten track to just to code email, or to call those organizations and see what, what they respond like. Yeah, I think if they can share the data and if it's not very sensitive, then I think, yeah, they would be willing to share those kind of things, if you are doing something good with it. Yeah. So that's what I am thinking.

[00:20:09] Ken: This episode is brought to you by Z by HP. HP's high compute, workstation-grade line of products and solutions. Z is specifically made for high performance data science solutions. And I personally use the ZBook Studio and the Z4 Workstation. I really love that Z workstations can come standard with Linux and they can be configured with the data science software stack. With the software stack, you can get right to work doing data science on day 1 without the overhead of having to completely reconfigure your new machine.

Now back to our show. That's awesome. That's a really cool idea for a hackathon is where everyone has the same dataset, but each team gets access to people within an organization to ask questions related to the domain. I would really like to see that that to me, I think would make a really interesting project.

Maybe it's a little bit longer than just a couple days. Maybe you make it a week or so, but if maybe we'll do something like that with a, with a couple of creators, what we'd love to get you involved? We'll we'll see, I have a lot of my plate this year, but maybe early next year, we'd be able to put something together along those lines, because I think that that creates a lot of value and it gives a lot of people the opportunity to potentially tackle a global problem.

[00:21:29] Thu: Yeah, totally, totally. I think, yeah. The person who post the dataset on Kaggle, I think their job should not end there. I think there should be a, like a, how to kind of committee. So, that, yeah, just to give some help to people like data scientists who are working with it. And yeah, I think that it would create a tremendous amount of value doing that.

[00:21:57] Ken: That's amazing. Are there any public datasets that you're really excited about that you think are pretty fascinating to work with? I think I have international criminal court here in my notes. Is that something that that's available to people?

[00:22:11] Thu: Oh yeah. Yeah, so it's another hackathon. I joined and so in the Netherlands, in Den Haag (The Hague) here, they have the the office of the International Criminal Court and they, for each case that they investigate, they usually collect a lot of witness statements about, like, for example, they might interview people who live near where terrorist attack happened. And so they just collect a lot of, lot of statements about yeah, how it happened, like, Did you see any suspicious people around and stuff like that. And so they asked us basically to use those statements and use NLP, or whatever have we have in our ... to like, make sense of it in a more effective way because they have to like the analysts have to spend days to read through those statements.

And usually it's very long, like probably like 5-10 pages for each of those. And so, yeah, it's a very interesting case. We couldn't really do a lot of good things we did, unfortunately, cuz it was a like, yeah. It's 2-day hackathon, but for most of that two days we were like trying to figure out. Okay, the data is a little bit messy. We don't have enough statements to work with, so we couldn't train any model. Using pre-train model is not really suitable because yeah, the witness statements are a little bit like you have very particular, very different. The English of some witnesses.

They are not really even good. So, we were really struggling with that case. But, yeah, if we could use some kind of like ... as yeah, ... or relationship extraction from that kind of dataset in an effective way that would bring so much more value and people could immediately see, Okay, what is the common kind of like object that people were talking about, and it will really speed up the understanding and they can see what they should focus on. Yeah, so it's another case, but we, we didn't really succeed and I guess, Well, it's a lesson learned that yeah. If we don't have enough data, we don't have enough resource to actually tackle that problem. It's also, yeah, like a false promise to use data science.

[00:24:56] Ken: Yeah. I think that it's important to know when data isn't, Oh, data science isn't gonna be as useful as you might think. I mean, we think of data science, a lot of the time as a cure all, and unfortunately, in a lot of organizations, that's just not true. We try to do the best we can, but sometimes machine learning solution isn't necessarily the best solution.

And so let's talk about maybe data analysts versus machine learning and working in across those domains. Can you tell me how you went from your junior data analyst position to the role that you're in now, and then I'd love to hear about maybe what's next regarding your professional career. What, what interests you the most?

[00:25:44] Thu: Yeah. Yeah. Good question. I for the question of like, the future I would talk a little bit more about that, but yeah, things are not clear cut to me. At all even if I get to this point so yeah, for my role as junior data analyst, I moved to my work at the moment that is yeah doing data science consultants for like Big 4 company.

So things sounds fancy and great but yeah, actually we. Yeah. I just do all kinds of different things and it's just like a very like a yeah, like it's not really a linear process, but more like, yeah, you just keep accumulating, accumulating those kind of technical skills and all these experience working. In a lot of different projects and you just gradually yeah, just act, yes. All of a sudden you feel like, Oh, you know yeah. You know, to do a lot of things Yeah. I went from learning basic art programming to doing some kind of like, yeah, learning Python, learning SQL, learning JavaScript, because I was also into doing data visualization and like a more customized kind of visualization.

And also like, yeah, I learned machine learning like a year later when I after I started my job taking some courses on course here, like a lot of courses actually. And just gradually, I just built more confidence. And now working in my job now, we just have so many projects. We can do so many things.

Yeah, in my whole career, I've probably only done two or three really machine learning projects. So two projects, two, yeah, creating two productionalized models and yeah, that is basically it's not a much, a lot of machine learning out of like probably 15-20 projects that I've been involved in throughout six years, five years, six years.

Yeah, so, in a sense, I feel like, yeah, machine learning sometimes they're a little bit overrated. Yeah, not like we always have such a good amount of data to do machine learning with. Sometimes it's a bit over queue. Yeah. The clients that I worked with have been like only, yeah, the bigger clients that have enough infrastructure to actually productionize a model.

But other smaller companies, usually they, yeah, they value more like the kind of like short term kind of insights. And so yeah. Data analytics, data yeah, dashboarding kind of like data analyst kind of role would probably be enough. Yeah. So it's yeah, I've been, I've worn so many hats in my career.

I've done so many different things like automation doing, yeah, data analyst kind of job, building visualization, dashboarding, also a little bit data engineering. And so in a way I feel like, yeah, I don't know what I'm doing now.

[00:29:31] Ken: Well it sounds like the consulting role is something that does give you a lot of opportunities to experiment and try new things and work around in a bunch of different types of projects. I think that that's pretty uncommon for a lot of the data work. You know, a more traditional data science role is doing. I mean, if you're at a very big company, you're probably working on one specific project a lot of the time, if you're in a maybe medium sized company, you'll probably have two, three projects a year, or it depends on the nature of the products. It might be more than that, but you're working largely with similar data within the same domain. How do you deal with jumping around to so many different subject domains? Very, very quickly.

[00:30:16] Thu: Mm. Yeah, yeah. That's that's a good one. I think, yeah, it ... down to like, you always have a little bit, yeah, in my job, you just have some pressure to like keep moving and once. Even like before this project and another project has already started and you just have to quickly, very quickly read up about it and to sometimes ask colleagues to give a crash course on, you know, like insurance or banking or whatsoever. I had up so many times or like HR very interesting stuff. Or yeah, like the whole healthcare system in the country, for example, and yeah, I think. It's just all about keep an open mind what you can learn and what yeah. Have a, like a confidence that you can learn anything like nothing is rocket science.

Yeah, even though there are some rocket science out there, but for the most part you can do, you can understand basic things. You can, yeah, just even on the surface, but if you just, if you have the basic rights, that is probably enough to do your, yeah, data science job and just really keep asking questions.

This is one of the things that I learned my, yeah, I learned my hard way sometimes I just, Okay. I thought. I can just read up about it myself. I can, you know, figure out like I can just go on Google and ask. Yeah. And to read about this and that, but actually it is really not an effective way to learn. Sometimes you really need to just step out the, your comfort zone and to really ask a difficult question, like tough question, even though.

Yeah, could be even dumb questions on the first glance and for my culture sometimes. Yeah. There's a little bit of taboo to ask really dumb question or to ask someone to repeat the, exactly the same thing they already said before, but it is really a crucial step, I think, to get things right. So many times I, yeah, I just kept well things inside me and I just, Okay. I can figure it out myself, but actually it is like a less efficient way to actually learn a new domain. Yeah.

[00:32:52] Ken: Yeah, I think with consulting, it it's particularly important because you don't always have access to your clients. So you have to make sure that you're asking the questions when you have access to them. So you're not, you know, interfering with their daily work. It's a little different. When you're on the job, if your client is someone who, you know, works with you every day, they're not going anywhere, there isn't necessarily like a super finite and you can, in some cases, just walk over and talk to them.

I don't know if you're on site a lot at all these days. I expect probably not, but it's possible. And I'm wondering, I mean, I think you talked about your culture. I'd love to learn more about that because I find that a lot of Asian cultures, they conflict very directly with more traditional consulting and management consulting. And how have you been able to navigate those two things as they relate to each other?

[00:33:46] Thu: Yeah. Yeah, the cultural thing when it comes to consulting it could be both. Yeah. It could be both good and bad in a sense. Yeah. In a way I've always been, yeah, like trying to be a good student my whole life. Working relatively hard and I'm willing to put in the work to make something good.

And I think from the culture perspective, I think it's a very good trait to have when it comes to like learning new skills and you just like you have a lot of patience to to learn it. And Yeah, you don't complain too much and things like that. And yes, just because our culture value that, that kind of like being hardworking, being patient and things like that and being humble.

But when it comes to yeah, understanding the business question and really, you know asking questions, at least from my experience, that cultural perspective, hasn't really been really helpful because if you are afraid of asking questions then, or you are feeling, you are afraid of losing your face because you ask a very, very stupid question in front of many, like many people or like important people from the clients, for example, then yeah. It also hinder.

You as a yeah, as a person who are doing this job, and yeah, from so far, yeah. I've learned a lot through my work that, yeah, this is something that is important to like, to be aware of, yeah, but it is a good start to already like to think about it and to like to say, I'm not, why I'm am I not asking this question? Is it because I am afraid some of something. Yeah, or because yeah, like I'm trying to hide my spirit stupidity or something like that. Yeah.

[00:35:55] Ken: Yeah. Well, I forget who told me this, they asked me what's worse, a stupid question or a stupid answer. And. A stupid answer. I think most people would agree is worse.

But if you don't ask the stupid question, the answer you give to a client will probably be a stupid answer and you don't wanna do that. That looks way worse on you than not asking the silly question and just leaving it and letting it like grow and fester into, into a bad solution that you've created.

So I think that if we think of it, that. You don't wanna over. I think the only way a question is bad is if you haven't done your homework or it's not well researched. If it's something that you can't necessarily research, because maybe it's jargon, maybe they're using an acronym that you're like, what the heck is this?

That's not a good question. That's, that's something that like, Okay, this is, yeah, exactly. You know, I'm interested. I mean, you've jumped around a lot. you, you work in a lot of different domains. Do you like your job? Do you like the domain? You know, there's something we've talked about offline about, you know, are we just following a trend into this domain or once we get there, we have to think about, you know, a lot of different things after we've been in the field for a while, you know, how do, how do you feel about that? How do you feel about this career?

[00:37:18] Thu: Yeah. I've been thinking about the same thing. Actually, yeah, when I first got this job, I was so happy. I was like riding my bike and like singing, Oh, I'm becoming a data analyst and stuff like that. Like that, I didn't really think about the trend and all this a lot and how it affected my decision.

But now as I am in this field for, for a few years now, I start thinking, yeah, harder about, yeah, it's, it's an interesting job. It is. But is it really indeed, like, do I really love it? Do I really see myself doing this for the next few years and am I making a good decision to stay instead doing something else?

Yeah, that is very, very. It's been kind of like bothering me for the past year or so or two years. Yeah, I think as I moved across different fields and doing so many things, wearing so many hats, I started to realize that, yeah, I'm probably not so much interested in doing kind of like.

Machine learning or like really like become a researcher in data science or anything like that, or into understanding the hard math under data science or all, all of that. I feel like I'm just someone who, who is relatively good at technical skill. And I have a little bit creativity with it. So now, also on my channel, I also feel like, yeah, what should I talk about?

Some people want to learn about machine learning. I feel like, yeah. I just want to do something fun with data science. So either solving a very interesting problem, like the lung cancer case or creating something kind of like yeah. Bit pretty out of yeah. A dataset or quitting some visualization or doing something fun with it.

So. Yeah, in a sense. I still, yeah, of course. I still love my job and that's why I'm still staying. And I think it's it's a very rewarding career. However, I also think, Well, in the next few years, probably I would think about, yeah, probably I could. Transition myself into something that is not very, very mainstream.

So maybe they assigned plus something else, like maybe doing even like data journalism or yeah. Data art or, yeah, just making, creating really unique portfolio projects that just inspire people to just yeah. Do something useful with data science and also yeah, just like really inspiring stuff.

Instead of like grinding on gaggle to increase the, like the performance of a model by 1% or 0.1% and things like that. It just doesn't click with me a lot. So yeah, I'm more interested in indeed like the cases, so like the actual problems and yeah, the creativity side of coding for data science. So yeah, I guess I would probably move my career to something else.

[00:41:04] Ken: Well, so how, you know, I'm interested in how you came to that conclusion about yourself, or, you know, obviously there was experimentation working on a bunch of different projects, working on a bunch of different datasets, but were there any other factors perhaps, you know, managers or other people in that process that have helped you come to those conclusions?

[00:41:26] Thu: Oh, yeah. Yeah. So it is because of some projects that I have done for clients actually, and Edward, my work. So I just created some like just a customized visualization and build some like, kind of like solutions that is like combining the, like the machine learning underneath and also like very kind of like fancy.

Fancy kind of like presentation in some kind of like web web app. And I just thought, Well, this is really cool. And I saw think, Well data science is just like the. Probably more the backend part and I'm feeling like, yeah, I want to build actual products and I the more tangible things that people actually see and interact with it and to see that like how people perceive these kind of insights on how people use those kind of, kind of tools, it just.

Makes me more, yeah, it just inspires me more. It just gives me more energy cuz I feel like, yeah, once I wake up and I create something fair, some like problems regarding like coding. It's just very interesting coding problems. Also sometimes when you work with like the kind of front end stuff and yeah.

Like at the same time you still have some, a little bit like data analytics, data, like machine learning connected to it. Yeah. I just think, Well, it's probably more interesting for me than yeah. Doing like hardcore statistics mathematics on the machine learning models. Yeah.

[00:43:18] Ken: Hope that makes sense. That's awesome. No. And is, is that sort of I guess different mindset. What also drew you to YouTube? I'm really interested to hear that story as well. So how did you get started?

[00:43:31] Thu: Yeah. Yeah, indeed. So the YouTube thing is it all started actually, yeah, not so long ago. Probably now, yeah, since may last year and back then I was watching your videos and looked video and I thought, Oh wow.

Also the Alex The Analyst. Yeah. Like you, guys, just kept popped up on my popping up on my home screen. So I couldn't help watching all of this. And I thought, Wow. This is really like, yeah. Some videos are so funny and so inspiring that I thought, Okay, I could do something like that. And back then I had very yeah, just a lot of time on my hand because I didn't really go anywhere in the islands.

It was like a hard lockdown the whole summer. And so I had just, Okay. I thought I could make videos, probably just my mom would watch it but I just wanted to try myself out and then see if I could create something out of it. Yeah. So indeed, like most of it. For most of the year, only my parents and my sisters in Vietnam would watch my videos and comment on it.

So it was really sweet because my parents also had absolutely no idea what I was talking about. Yeah. So I thought, wow, gradually just having more and more people who are interested and just thought just gave me good feedback and. And I could also do a lot of things regarding like, just as a creative outlet for myself.

Even in data science, it's a very like pro kind of like sometimes a bit boring or dry topics. But I just kept thinking about, Okay, how I can make it more interesting. how to how to visualize this thing or how to use graphics, to explain some kind of topics. And that really like intrigued me and I just thought, Okay, this is really cool.

I also love like doing video. So that is like a good combination for me to kill my time. But actually now it's, it killed, it kills more of my time than I would, I would have hoped for. Yeah, so that's a, that's a thing. And yeah, back then I had a, like a kind of like ambitions to combine painting, or like what, like visual art with with like explaining hard topic.

So I just, cuz I'm also doing painting in my own time and I thought, Okay, it'll be like a very unique science channel to be able to do that. and so far I haven't been to do that, but sometimes I do talk about like painting and. On my channel and some videos it is very hard to actually marry the two things like your hobby and the data science.

And as you do in the channel, I think you have really succeeded in doing that and find the right audience who are interested in sport and data science at the same time. But for me, like, Painting and art and their science. I feel like it's a very hard marriage to make, but yeah, so far I've been trying some like stuff on my channel and some people really liked it. So I think I'm doing something good there in terms of like using some more like editing some creative elements on, on the video. Yeah, so that is how I got into this.

[00:47:10] Ken: That's awesome. how long have you been painting for?

[00:47:15] Thu: Around three, four years, I believe. Yeah. Mostly, yeah. I'm not very good at so I tried even tried to sell my paintings on, on Etsy and the only person, or like two people who bought it. One is the mother of my ex-boyfriend and the other one is someone in the us, I don't know. So at least I have some comfort. That's like, I could think of myself. Okay. My painting is not that bad.

[00:47:47] Ken: That's amazing. Well, yeah, it's, it's really impressive. How high production quality your videos are. So something Luke and I talk about is yeah. How you sort of came on early and you were able to make really high quality videos very quickly. How did you pick up the editing skills, the camera skills and those types of things to be able to create these videos? I'm always interested, obviously in skill acquisition. And I'm wondering what your process was there.

[00:48:19] Thu: Yeah. Yeah, a lot of people also, my colleagues also asked me about, Okay, how I made videos. Yeah, I actually, I never I've. I had never used a, like a digital camera before, before I met this channel. I just bought the camera just to like practice and also on zero editing skill, but actually. Yeah. I just felt like, yeah, I just feel like I pay attention to the things that I want to learn. Like when, once I want to learn something, if I watch like a two hour video, I could, would watch it like the whole video.

It's just like my attention span somehow, just like really like, just like a genius. And it's just like, I just kept, watching, watching a lot of videos, tutorials and just do it myself. Actually, my. 20 videos on the channel was added on my iPad, on the iPad pro. So, it was really nothing fancy. But yeah, there's just so many things that you could do with what you have.

And I think it's not about yeah. Having the like fancy gear and software, but more like yeah. How you just think about putting yourself in the, in your, in your viewers shoes and see, Okay. What could be interesting to put on here? Of course, yeah, the lighting and the audio, everything. I've only.

Yeah. Figured out the, like the audio completely very recently. And Luke has been like, kind of like, Oh, you need to improve your audio and like, Okay, I will try to do that. Yeah. So it's very, a very messy process for me. I'm yeah, way more messy than, yeah, seeing looks process when he, where it just like, have like a full script of like everything, like thumb new and title and everything.

I, yeah, like I've always been very messy, I guess. I just write a big chunk of text, like with no, really like no paragraph and just like try to chunk everything together. Title and thumbnails. I'm very bad at that still. That's why it hasn't really grown very fast. Yeah, so I would say, yeah, just like being curious and willing to learn, like seeing some people who do something that you like, and you just remember it and you just try to find out how they did that. Just watch a lot of YouTube videos as well, because that's a I think that's, the input is really important for the output as well.

[00:51:07] Ken: That's awesome. I think everyone, I've learned from meeting a lot of the creators that everyone sort of has their own process. And it's not about copying someone else's process. It's about refining, what works for you. And of course you can borrow from other people, but the way Luke does something, the way Tina does something, the way you do something is not gonna be necessarily what works best for me. And there's some sort of, you know, just like you're finding your way through the data domain and figuring out what you like and what you don't like.

Almost everything in the world is like that every, every process is okay, how do I approach this? What works the best for me? How do these other people approach it? Can I borrow something from how they do it? Yes. Great. No, that's okay. And move on and continue to learn. So I really like the over the overarching sort of message associated with that. Those, those are all the questions that I had for you. Do you have any additional thoughts? Where can, where can people reach you and learn more about you?

[00:52:09] Thu: Yeah. I think like one of the things that I forgot to share as well, like what I learned from actually yeah. Luke told me like long ago was that you should try to be, yeah, yeah. Be different is better than better. And I think it's a very, very good advice. Not only for like, yeah, for us as like content creators, but also in our data science domain or field, I think it's really important to really know your strengths and know your yeah. What is the unique combination of your skills and domain knowledge that could potentially help you progress into this field?

And so I think, yeah, like being different. Is very, very important in the word today. Being a little bit better, probably doesn't pay off so much, but being different and make a good choice based on your insights is yeah. So even more important and sometimes, yeah, choice is more important than effort as well. From school.

Yeah. We all learn, like you have to make a, you put a lot of effort into like studying and to follow this path and be your best student and things like that. But once we get into. The work the workflow you get into the like, yeah, figuring out what you want to do then, yeah, having insight and make a good choice is even more important than putting a lot of effort.

Sometimes yeah, make a good choice in terms of like talking to person reaching out to this person or like yeah. Get out of your way to get involved in some particular project that you always want to do. And things like that. I think it pays off much more than trying yourself to make a lot of effort and just hope that something will happen.

Yeah, so, being different and make good choice. I think that, I think that are my two lessons that I learned not only from YouTube, but also from my work as well. There are tons of consultants. There are science consultants out there. How can you distinguish yourself? From, yeah, like, yeah, dozens of colleagues around you who are doing exactly the same thing and not, not if not even better.

So yeah, like from that perspective, like we really need to stand out in our own way. Just like a creator actually yeah. To get the right attention yeah. To be able to progress and to develop. Yeah. So that is a very, very hard part to. to figure out, but also like very fun.

[00:55:18] Ken: I really like that. I think different usually does mean better too, in some sense where, you know, if you're different, you're playing in a smaller circle, you want to continually make smaller playing fields for yourself to succeed. So if I'm a data scientist, if I'm slightly better than the average data scientist, that's not gonna mean really much.

But if I'm a data science, for example, YouTuber, right. Where there's. Maybe 150, maybe 200. I really don't have account if I'm, if I'm slightly better there, that means that a lot more people are watching my content. A lot more people are interested, whatever it might be. And that's also sort of a collaborative space, maybe not the best, the best metaphor, but the smaller circles that you can associate yourself with.

The more you differentiate there, the more likely you're gonna stand out because there's just less people to stand out against. And I think luke and your description there sort of nailed that on the head.

[00:56:15] Thu: Yeah. Yeah, absolutely, absolutely. Yeah. And that is a very smart strategy actually. Yeah, just to navigate around this field and also like, yeah, just to become something that you actually want to become instead of like fitting yourself in this kind of like huge category of you know, titles of data, scientists, analysts, or whatever.

That doesn't make it doesn't make us feeling feel fulfilled. And also, yeah, it doesn't really like exercise all of the potential that we that we actually have. And so, yeah, absolutely agree.

[00:57:02] Ken: Amazing stuff. Thank you so much through everyone watching. Definitely check her YouTube channel out. I think you're also on LinkedIn as well. So I'll put all of those links in the description.

[00:57:12] Thu: Yeah, thank you so much, Ken for having me today, it was really fun. Chatting about all of these topics that we don't usually talk about. Yeah, thanks a lot.

[00:57:23] Ken: Amazing.

10 views0 comments