open-OH-2020-09-18.mp3 [00:00:00] What's up, everybody, what's going on? How's it going, man? Welcome to Open Office Hours. This the @theartistsofdatascience open office hours. We're just gonna get started here in a couple of minutes. So sit tight and we'll get going. All right. Got some people coming in right on. So check it out, guys. [00:00:21] What I'm going to do right now is going to pull up a really insightful post by my friend Giuseppi on a car. So I thought it was really interesting and insightful were his definition of a mentor is. And I think that it bears sharing with everybody here. He's talking about what are the right and wrong behaviors of a Data science mentor. I 100 percent agree with every one of these points. We're neither a professor nor a master. [00:00:52] We're more like a colleague or a friend that's here to support you, but not a repository of all knowledge. Um, and I definitely agree with Giuseppe. Giuseppe say that he's not a master professor. Then shoot. By no means am I won. Also are not a substitute for books, stack, overflow or whatever help that you can find for yourself autonomously. So I like what he says here about what you achieve without any direct help is a new summit that you've reached. So that's pretty awesome. We're here to just share our experiences, not necessarily all our knowledge. And I'm definitely not going to be influencing any important decisions, but I will talk through them with you and I will be asking questions. Just as you know, you guys have questions for me. [00:01:41] And I don't understand the question of the throw right back at you guys. Um, so. Yeah. That being said, I'd put everybody on mute, all right, so if you have a question, what I would like you to do is actually type out your question and put it into the chat, and that is going to hold your place in line. And that way I could also field questions. So please be as clear with the questions as possible. And I'm only going to be answering questions that are put into the chat. So go ahead, take a minute or to think about your questions, what is it that I could assist you with and then we will get the show on the road. I will do this since I was here first. Let's go ahead and just get the ball rolling with you. Everybody else, please. If you have a question, go ahead. Put it into the chat. I hope I'm is over for Caridi. [00:02:42] Yeah, it is. Alfre. Alfre. And how are you doing? [00:02:45] I'm doing good. Yeah. And Michael are polishing it. My pleasure, man. I'm going to help you. Yeah. Yeah. [00:02:52] I just I feel I'm actually like I'm I'm I'm trying to break into the Data science because my, my, my, my, my current garbage is mostly like in the social science. I'm doing econometrics and this type of like causal inference work and then try to break into like well I got into the science bit and yeah. So I just like I been attending webinar, I've been like attending like a lot of like even to kind of like put in my, my, my, my, my perspective. [00:03:24] And I just like it. I just feel like it. Since you are hosting this and I, I feel like this maybe this will be a very good, very good opportunity for me to get to speak to you. [00:03:36] And I just have like one question because I think, like, you have your own in your pocket and then you also have, like, a lot of experience in this field. What I want to come and what are the common mistakes of, like, aspiring Data high in this that you have to observe. And then, like you say, like people should not be making this mistake. It's something that you observe across your like your your your your experience. So I think like maybe like you can share it with us and then so that we don't we don't repeat those mistakes. [00:04:08] It's a bit of a broad question. Common mistakes in any particular domain or any particular application, a mistake in what respect can be? [00:04:18] I don't know. [00:04:19] But maybe it's more like like I like in networking, maybe like networking with like I would fellow Data. I would like job search, that kind of stuff. [00:04:30] And also maybe it can also be like in technical, maybe you have work with, with a junior or a junior, have a personal and then they make like this this mistake the right away. [00:04:41] I don't know if it can be anything. [00:04:43] Ok, so let's start with the with the networking and um go ahead, keep yourself unneeded offer. And by the way, everybody else, please, if you guys have questions, type them out, put them into the chat. That way I can get to your question intern. But when it comes to networking. So I've I've had I've done talks across the city here and I've been at other networking events myself. [00:05:09] And this happens every time there's at least two or three people who do this who literally follow you around with a resume in hand and just ask if you have any openings, job openings, and they'll just like follow you around for the rest of the the conference. Don't do that. Don't be that guy. Right. [00:05:32] So I talk about this a lot, actually. The podcast I released earlier this week on Monday with Alison Grade, and she had a wonderful five step approach for networking. And I'm actually gonna be writing about this pretty soon. But let me pull it up and I'll share with you what I've written from from this. So give me one second, because I absolutely love her framework and will pull this up. Yeah, because networking is a tricky thing. All right, cool. So I'm just pulling up some grades book here. I can find it. It is. [00:06:16] All right, cool. [00:06:18] Yeah. So you guys enjoy this one and it's super simple. It's super easy to follow framework and it's pretty much this. Right. So it's a series of five questions you can ask when you're at a networking type of event. And then based on the responses that a person gives and you know what to do next. Right. So first things easy, right? If you meet somebody, just ask where you based. How does this manifest itself in a Data science conference, you know, oh, hey, where are you visiting from or are you from out of town? I say simple open, like she says, to get the conversation doing, obviously. What do you do if you're out of Data Science Conference, Data science networking event? That might be a bit of a redundant question to ask, but I would instead ask, oh, hey, what type of problems are you working on? What aspect of Data science are you interested in? Data engineering, machine learning, ingenuity of scientists. An analyst, right, and this how does that work exactly, how does that play itself out for Data science events? Well, instead of how does that work exactly? You can say, oh, well, how do you how do you enjoy that? How do you like that? What challenges are facing in that type of position? Right. And then ask whether they're. And then the fifth step probably doesn't count for Data scientist, who can I introduce you to? [00:07:43] Well, maybe it might. It might be. [00:07:45] I just general five step framework, I think is a really good way to just not be awkward in a networking event. And it's really just boils down to just being interested about the person that you're talking to. OK, yes, just being genuinely interested in the work they do, because I remember like the goal of networking is not to ultimately get you a job like that might be a byproduct of networking, but the purpose of networking is to build a relationship with the person. [00:08:22] Yes, that's fair, that's that's really helpful. [00:08:27] Ok, awesome. And then you had a second question on top of that. [00:08:32] And I think I in my mind, most main concern is it's always like an unwieldy how to navigate the job market because like, I'm I'm currently based in the US and I'm an international. And then, like jobs, it is challenging even in normal times. So now it's like now there is a cobwebbed. Yeah. I think like my my second question, I think God, let me just dispel that myth real quick. [00:08:55] So um, so you know, my other mentorship platform that I'm a part of, the one that's hosted by Karl McHugh is Data sunscreen job. I'm the principal mentor there. And this summer I've we've had like 10 or 15 students land jobs. And every week my students are getting interviews and every week they're coming in and sharing about all these different opportunities that they're. Having come up and I think it's just the result of taking action, so if right now you aren't having any. Opportunities pop up, assess whether or not you're actually taking action. All right, so are you going out there and actually submitting applications or are you not even searching for a job? Right. And if you do go out there and submit an application, are you following up with them? Are you reaching out to people on LinkedIn? Right. Because you can't simply just submit your resume and hope and pray that it gets picked up by somebody on the other end. There's still work to do in between that. So it's all about creating opportunities for yourself in that respect. [00:10:11] Yes, yes, thank you. [00:10:12] Yeah, I think that there are some other questions on the maybe. [00:10:16] Yeah, you know, we had another question that was really good, too. I just can't remember what it was. [00:10:22] No, I think yeah, I think the questions like a common mistake, especially on the court, on the deadlocking. My my second question is actually like how to network effectively. But I think you can address that question like that on. OK. [00:10:36] Yeah. Definitely feel free to hang out and thanks for asking. Yeah. Gunjan just saying thank you for starting this. Hey man. It's my pleasure. So as part of Data says dream job, I have office hours like six hours a week. So I wanted to help more people. So I figured if I just carve out maybe an hour of my time and try to give back to the community, help people out, I'd be happy to, you know, keep in mind. By no means am I an expert in anything. I'm just. I'm just a dude who works as a data scientist. That's pretty much it. So thank you for being here and thank you for. [00:11:20] Entrusting me with your time. Sonia Mowbray has a very. Detailed question, um, Sasanian is a data scientist with three years of hands on work experience in the field of machine learning, he or she is looking to transition to NLP and computer vision. They read quite a bit about both fields through courses and books and have recently started working on a couple of NLP projects at their work, at their current job. [00:11:55] And they're taking on freelancing jobs in the same domain. Well, you should check out the episode I released on Monday, um, about freelancing with housing grade. [00:12:08] So you got a couple of questions related to the same. Should I gain expertize in just one of these two areas, NLP Invision. What would be helpful to gain a knowledge and hands on experience in both quite interested in further studies, so it would be beneficial from a career standpoint to go for Masters? So would it be beneficial for Masters or pursue Masters through an online university and see if the advantage of not compromising on your work? All right, um. Let's talk about this, so you're still wondering if you have to pick between NLP and computer vision, why is it that you think that you have to pick between the two up here, read this and say I'm walking in from India and I really appreciate you picking out paint at this session? [00:13:05] Ais. Yeah, so they're the reason that I think that I may have to choose I know one of the boys because of the rapid pace with which both the COVID got me and day in and day out, new models and new architectures are coming in. And what the key is, and it's really hard for one to keep up, keep a piece, but news at the same time, as well as maintain quite an in-depth knowledge about both of fans. All right. So real quick. [00:13:35] So let me ask you where it is that you see yourself then. OK, so you're saying both fields are very rapid, they're moving very quick and they're changing at such rapid pace that you're worried that you won't be able to keep up with one or the other. So where are you trying to be? Are you trying to remain an individual contributor type role or are you trying to move up through leadership management? Like where do you see yourself in the next few years, like in terms of the hierarchy? Where do you want to be? [00:14:04] So so given my current work experience in the two goes on for the next couple of years, I guess I see myself as an individual contributor or that I definitely see myself moving up the corporate ladder and transitioning into certain management and of within the same team. All right, great. [00:14:24] So if you let's take this one step back now. Let's think about. The principles for NLP and computer vision, do you think fundamentally the way you saw the NLP problem or the way you solve a computer vision problem is going to change? Is the principles of your approach or is it only the tools that change? [00:14:48] I need to eat it plain. I mean, much I can. It's the cleansing part remains the same NLP model meaning. [00:15:02] And if you're confident in your ability to pick up new things quickly if you need to. [00:15:09] Yes, definitely. [00:15:10] So it's not like if you take a NLP job and you're using one particular tool that overnight assumes a new particular tool, rules out that whatever organization you're in is rapidly going to adopt that tool, or do you think they'll stay with whatever solutions they have now and either adopt the new tool or not adopt the new tool? [00:15:33] So this is my current work experience in looking for an. I see that a bit slow moving forward with getting that are coming into this year. But I expect there will be certain inertia when we go into that sort of thing. [00:15:55] And that same inertia is present in many organizations. Right. So even though there might be a seemingly rapid pace in the way the tools change, the way you solve the problem itself doesn't really change much. So when you look at it from that perspective, that it doesn't become as mutually exclusive as you're painting it now. Right. So if you have a passing interest in both, I would say do what's more delightful to you. Do you like do you enjoy NLP more than computer vision into a computer vision, more than NLP? Or do you like them both? Right, because they don't have to be mutually exclusive. Um. And if you're looking to go up to a managerial role in a couple of years, then you don't have to worry about keeping up with the new stuff because that's what your, um, your team would be responsible for. So in that respect, I would say if you enjoy both of them, pursue both of them. But like I, I don't think this is necessarily like a binary, you know, problem. You can't you don't necessarily need to pick just one and pursue that one, because even if we abstract away one level further and ignore the NLP or computer vision aspect of it, it's still a machine learning problem. And generally that pipeline is still going to be the same at that higher abstract level. If you just move up a couple of levels, that makes sense to me. So, so, so I guess my response to that. So your question is, would it be helpful to gain knowledge and hands on experience? And both. I think the more diversified you can make your skill set, the more invaluable you make yourself as an employee. So rather than I'm a big fan of generalizing, not specializing. So I think if I was as smart as you were, I would take both of them, like I would study up on both of them and have both of those skills in my town stack. [00:18:04] So that is making it that. Yeah, yeah. [00:18:09] And then the other question here is interesting. In further studies, awesome. Would it be beneficial from a career standpoint to go for master's or pursue masters through a online university, since it gives you the advantage of not compromising on work? Yes. So if you're genuinely interested in learning more and that's like you're doing it just for the sake of self education because you thoroughly enjoy it by all means, like like go for it and go for masters, like it's all about educating yourself and getting better. And you're wondering if you should do it online or in person, um, whatever's going to fit your lifestyle best. So if you are like if you want to continue working and you don't want to relocate for school, then yeah, definitely do it online. If you don't if you can forgo a couple of years of no income, if you can go through that, then go to in-person university. But I don't know how person universities are going to be nowadays, but, um. But yeah. So I mean, that's kind of a it's not really an answer, but I would say for that question, think about what it is that you value. So. So talk to me here. What is it that it's like. Do you have enough money stacked up for a few years that if, if you wanted to quit working and go back to school, that you'd be OK? [00:19:37] Yeah. So I mean, again, just for my future education, it is hard to keep that moving forward. [00:19:51] So do you think that not having a master's, is that causing any impediment to your career advancement currently? [00:20:00] That's a great question. So since I've been planning to quit my job, to make it somewhere I have been applying for a couple of other jobs is very often the same thing as I see many of the companies misstatements to us as a qualification on their job, although I think the Masters. But that's an MBA and I would want to give that many, many of the companies are looking for a masters in it, you know, in a statistical data. That is why I'm contemplating a master's. Mm hmm. [00:20:46] Yeah. So just like I'm not sure what the what the job market is like in India, but like, let's say if this was kind of like, you know, Canada or the states, I'm based in Canada. Um, if you were if you've got a significant number of years of work experience, like you mentioned, you got three years of work experience and you're quick to pick up things and you've got a few solid projects that people can look at that are artifacts that demonstrate your technical capability. Um, then you having a masters in that field probably wouldn't be as much of a deterrent, because as you go through the interview process, if you're able to answer all their questions. Sufficiently and provide really solid answers, and you're able to demonstrate that even though you don't have masters in that particular field, that you can hold your water, hold your own, then it's all good. Um, but again, I'm not sure if it's different in India, if they just don't even look at people who don't have masters in that particular field. So, um, if it was. [00:21:53] Let's go on. Go on. Sorry. [00:21:55] Yeah. So that's the point. More supportive about because many of the I thought there would be a qualification for LinkedIn is getting a masters in the computing field because I mean just like any interviews. [00:22:17] Yeah. So I'll just say straight from my perspective, I was in your shoes given what everything you've told me I would. I would not want to stop working, like I would still want to work, because that experience working in the industry like that, super invaluable. Like it's really, really like there's no substitute for it. Right. So I would keep working, but then maybe pursue an online master's degree in my free time. That way, it's like I don't have to not have income and I don't have to be detached from the industry. [00:22:53] Oh yeah. [00:22:56] So hopefully that's hopefully that's helpful, man. [00:22:59] Yeah, that that's definitely helpful. So thanks. Thanks, Chairman. [00:23:06] Thanks for the question. [00:23:06] So let's we got, I think, a sharp. [00:23:12] Going into a deep analytics interview for the first time, right on what are some of the what are some of the things that stand out and compensate perhaps for a lack of practical experience? Do you have any projects? [00:23:27] Yeah. So just to I just wanted to brief because if I gave you a lot of details, I think that my but essentially, just to summarize, I'm from London, U.K. and I basically have one years experience. I would say pretty much OK. So like going from let's say from just data collection, organizing surveys at a basic level to then progressing through like 80. I don't know if you have with agencies is to marketing. I'm not too familiar with that. But essentially it's just like collecting different metrics, like whatever it takes. I say like, you know, you have sessions, e-commerce revenue, just doing that type of thing. And then another one is that web scraping in a sense to like getting information from Data. So, I mean, online RSS essentially is using Python to get information. So, yeah, just a combination of Python. And so you can say over the last year or so, not just from practical as well, but mainly from projects, because I've been learning those too, for the last two years that make sense. But essentially, the real practical haven't really done I've obviously done a bit more python, but as you say, but if that just clarifies a bit of my background, roughly in a sense, I have more project experience, like from my personal projects rather than practical experience. [00:24:56] Yeah, I think, honestly, the personal projects and they can substitute for practical experience if they're really well done. Personal projects. Right. So because I've seen personal projects that are just like really haphazardly and not really put together that well. But if you've got a personal project that is a deliverable that you would be proud to deliver for your full time job, like if it's that level of quality, then you could substitute for a lack of that practical experience for sure. [00:25:26] Um, I'm not sure if that answer your question, but, um. [00:25:30] So, yeah, just elaborate a bit more fair. And I did. I don't know if you'd say yeah. Yeah. So that's basically what I did essentially for this project and it was in at help and services. And yeah, I sharpened up all those online courses. [00:25:46] They're great to learn from and it's great that they provide you with like blueprints for how to create a project of your own. But I think the best learning comes from creating a project that is really original to you based on your interest rate. So you said you mentioned you had experience doing web analytics, digital analytics. [00:26:08] Yeah, right. [00:26:09] Yeah. So if there's any way that you can source data regarding Web analytics, uh, and and pull that data from the Internet, maybe through some API or whatever, and do some Data cleaning it, they do some data manipulation, some aggregate organize and then like make it a real original project, come up with a hypothesis, test your hypothesis, present your findings. Right. Um, and. Have what you use, what you've learned in those online courses as a blueprint or a template of knowledge, and just to have that project as a capstone, I guess, this week called the States and Canada Capstone Project that takes all the skills that you learn in the U.S. course and you're applying it to a original brand new problem statement. [00:27:00] And so the second part is just elaborate on specifically the lack of practical, because I've been doing a lot of so I've been getting quite a few feedbacks and from interviews and stuff like first stage or seconds, Sarkozy sometimes and some of the feedback is generic, like when I ask, it's just like, oh, compared to other candidates, you lacked some specific experience or that with the job description. Let's say one of them was Data Data quality analytics, for example. And essentially you talked you and you are really excited about the Data engineering aspects. We are more excited about NLP and AIs, which I think is a dumb excuse anyway. But that's what he said. [00:27:46] Yeah, well, I mean first first things first and then just make sure that you're applying for the roles that are genuinely exciting to you. So yeah, limit the scope of roles that are genuinely exciting to you. And then when you do get a callback to see you apply online, you get a callback for this particular job. [00:28:02] Then the next step would be to treat that job as so that posting job posting as if it was like a syllabus. Right. And what I mean by treelike is a syllabus I go through that I take like a frickin highlighter and highlight every combination of words that don't make sense to you, like, oh, I don't know what with this particular thing meanspirited with that thing, start sort of looking them up and started developing mental models for yourself so that you can research about it, get a little bit educated on that. So when the question comes up, you're ready to talk about it. And even if you haven't done that particular thing. [00:28:37] Explicitly, through your work, you can at least say, you know, what happened during work, but I will tell you what I do know about it and shout it out. And if I was to have to do this thing at work, this is how I would go about doing it, right? [00:28:51] Yeah. [00:28:53] That's like probably the biggest thing you do is literally like treat that job posting like it's a syllabus for some exam straight up. And then when it comes back to time to like ask for feedback, we say, unfortunately, if you do get rejected, like. Just make sure you're asking feedback in a direct way and be just. [00:29:16] Overly grateful for the opportunity. Thank you so much for taking time out of your schedule to to meet with me. I really appreciate it. Totally understand that there's not a fit between my current skill set, which you guys are looking for. But I do appreciate the time that you guys took to speak to me. I look forward to hearing about all the awesome things you guys are going to do in the future. Would you be able to share with me what I could do to become a better candidate in the future? [00:29:40] Okay, I say fair enough. All right. Let me tell you. Just say the last five. Yeah, I think that's a good point. But I think I might just say this is worked for other people just in general. One valid point was actually that I was I wasn't pausing enough. [00:29:57] And, you know, sometimes when he says, OK, so tell me, in this situation where a negative experience and then I always speak very rapidly sometimes to the response, and I think it was valid because they said that you going forward, you should take a pause and breathe. And so I think that's fair enough. That was a valid point. And so ever since, like two months, I've been doing that. Have you heard of, like, the star format? Yeah, yeah. That's what we used in the UK. [00:30:22] Yeah. So like the other big thing is, you know, look through the commonly asked behavioral questions, you can find them anywhere and literally just. Come up with the word document for each question like typo s, underline it and then a few bullet points t bullet points and then just practice telling the story. And how you can practice on the story is there's a situation when dot, dot, dot, dot. My task was to do dot, dot, dot, dot, dot. The actions I took was dot, dot, dot, dot, dot. And as a result, dot, dot, dot, dot, dot, dot, and bake that into how you respond to questions and just use this as a guidepost. Um, we literally just baked that in Trancer every time. And that one single track that we stay on track that we like. OK, well situation well above talking forever. Going to fucking said tasket. Oh shit I but my task was to and then you keep talking. Right. And then if you don't hear yourself say action or analysis or whatever then it's a cue. [00:31:29] Oh shit. I tell them about this. Right. Some kind of a mental wave away stranger on. [00:31:34] Hey you know, sorry about that guys. Yeah. [00:31:36] I can hear you know, I hit the chord, I hit the quarter my microphone and and that caused it to blink out. My apologies. What was the last thing you heard me say? [00:31:51] It was about like if you missed a point on a or then go back to it, yeah, it's literally just like a road map to keep yourself on point, right. [00:31:59] So by breaking into, like, those words, situation, task, action, result in every story that you tell, then you keep yourself on track and you're just getting right to the point without losing anybody in this stream of words. [00:32:15] So going back to that, just quickly, how would you advise that, let's say like once a weekend or whatever, would you so. [00:32:24] This would sound really weird, but share my story with anyway, so, I mean, I haven't really been in the interview game for a while, but when I was in the interview game and by the end of the game, I mean, like like I had multiple interviews a week just going insane. I would just practice interview questions in my head while I was going on a walk. [00:32:47] Right. [00:32:48] Okay. Fair enough. So what can help is, first of all, look up the people on LinkedIn, right. SQL find videos of them so you can figure out what they sound like, what their mannerisms are. Drive by the place that you going to work at. [00:33:04] Imagine what it would look like inside of there and then just take a. [00:33:09] A few minutes, like 20, 30 minutes, just sit back and just imagine yourself in that interview and imagine them asking you that question, OK? If they ask me this question, this I'm going to respond. And if I respond this way, they're going to ask this. So then I'm going to say this. [00:33:26] And then if that's not satisfying enough, then I'm going to come back and give this example. [00:33:33] And then they're going to say, you know, have that imaginary conversation in your head with the person and walk through that. Before we actually get to the interview and you do that enough times, it's kind of like there's a bunch of there's a cognitive trick, but your brain's not going to know the difference. So that when you're there some to feel like you're already been there before. [00:33:59] Yeah, fair enough, I mean, there's a good chance thank you very much for sharing those. [00:34:04] Yeah, no problem. All right, so you got Kunjin. Let me read this real quick, Gunjan, to say he's got a year and a half of experience as data analyst from his home country, Gunton Shamos, probably from Punjab. I was working majorly with Excel, Oracle, BI and Tablo, automating dashboards, preparing reports KPIs. After that, you shift to banking, which is not really a data analysis. And now you're in Canada. Welcome to Canada, my friend. And once again want go back to Data analysis. Do you have a chance? What are the broad fields or job positions in Data analysis, how shall work towards it? Do you have a chance? Absolutely, man. Everybody has a chance. You got a chance. You got a chance. Everybody has a chance. What are the broad fields are job postings in Data analysis? This is a great question. So. [00:35:00] So we're going to say something. [00:35:03] No, go on. OK. So I think what people tend to do is when they are looking for jobs, they'll just type in like data scientist or data analyst. But on LinkedIn, you could actually search for the particular skill instead of the actual job title, because there's a bunch of job titles out there that that aren't even they don't even say data science or data analyst. [00:35:26] But when you look at the job description like, oh, well, clearly that's what this is. Right, exactly. So I think it was. Do you know Eric Weber from LinkedIn? Sorry, do you know who Eric Weber is from LinkedIn? No, let's. The pull up real quick, so he had like this awesome list of, like Data science related roles that didn't involve the name Data Science and the job title. So let's pull that up and definitely follow Eric. He's got a lot of amazing. My. What do you see on my screen right now? [00:36:02] It's loading. It's. Yeah. Your schedule. OK, I can see it now. Yep. [00:36:08] Awesome. So where in Canada are you. [00:36:13] I'm in Vancouver. Oh, very nice. [00:36:15] Yes. So my husband was there in Winnipeg, so that's where I live. [00:36:22] I know I lived there for three months. I was there in Winnipeg as well. Then I shifted to Vancouver in the winter. [00:36:31] I know. In the summers, luckily. Yeah, winter is nice. I like winter, but. [00:36:37] But I saw the winters in Winnipeg for sure. I came there in February so it was quite wide and temperature was minus 15, 20. [00:36:48] So that's like the heart of the AIs. So for everybody listening on the podcast. So this, this episode, this is going to be released as a podcast episode just heads up. [00:36:58] What we're doing right now is we're going to Eric Webers LinkedIn and Eric Weber very recently posted a really insightful LinkedIn post where he was talking about. You could see here just the quality of all his posts are just amazing. A lot of great insights following him. Please do. But he had this really cool post about all these Data related jobs, but none of them had the title data scientist. And I'm hoping that we can easily pull that up just so we can kind of give you an idea. And if that doesn't work out, then we're going to do something else. That was it that long posted this. Sorry, go on. [00:37:33] I have been following you and even he puts a lot of interesting articles out there, so. [00:37:42] Yeah, definitely, yeah. One hundred percent vote for Kyle. He is my good friend. Hopefully you guys follow me as well. Um, my numbers on my engagement aren't that high, but hey, it is what it is so I can't find his post because I guess it must have been a really long time ago. So we'll scroll through just a little bit more. Just a little bit more. [00:38:09] I wanted to know this as well, like, so I'll give you a brief background about myself, I have done my master's in economics, so I had statistics and maps all that and as my subjects. And soon after that, I worked as a student for one one, one one and a half year there. I would say that it was just like learning ABC. For me, it was just a starting point. And I, I actually didn't even knew anything about it, so I was just new to it. So I learned a lot of excel there. Then we were working on article by and then w and most of it was data visualization, then preparing Diebold, automating the automatic reports and all that. [00:39:01] Have you kept up having brushed up on the scale. Have you done any side projects because the biggest, the biggest benefit for you would be to kind of practice those skills again and be able to demonstrate those in the form of projects. So that's the biggest thing. [00:39:17] So I have been doing because after that I shifted to banking, but I still have that love for data analysis and data. And it was not that easy for me to shift back to Data so far just to keep myself up and keep trying. I did Babloo course online that I did and advanced Excel course also online that I, I am doing SQL as well. And I have recently got to know about this as a website called Meikle Makler Monday. Yeah. So they put in the data and you can use it. It's very and it may small Data but they, but you can see how people are doing so much with that data. So I have started doing that. So that is it. That is where I have reached it now. [00:40:12] So I don't I get confused at how how should my part be and how interview is like a distant dream for me right now. [00:40:20] So I, I just want to I would, I would say apply anyway is like always be applying, always be applying because you don't want to shut yourself off from opportunities just because you don't think you're ready. Right. So you don't think you're ready. You should still apply. So go for, go for it. Like there's nothing stopping you from doing that. But the most important things is building out projects and namely building projects that you find interesting that are genuinely interesting to you. So make sure that you identify. [00:40:54] Yeah, because nowadays there's so much so much of information there and so many things going on. So if you are suggesting that I should start with doing some projects, so from there, should I do it from where should I get it right about to get there. [00:41:11] So what kind of industry do you want to work in? Do you want to work in health care, doing work in e-commerce? Do you want to work in manufacturing? Do you want to work in. [00:41:22] I have started looking out and I was I want that if I could use my finance knowledge as well. So if I could get some profile of my finance knowledge plus my Data could be used, that would be best for me. But as of now, I have I, I actually don't have that much of knowledge right now, but I am trying to find something like that because I am from a commercial and a economics background. So I think that will be best suited for me and that would actually be a be of interest for me. [00:42:01] Yeah. So what am I do to show you how you can try to find a project for yourself and maybe some Data for a project. So anybody listening right now on the podcast, this is all going to be put on YouTube. So if you guys want, you can go on YouTube and see what I'm doing. But pretty much I'm going to open up a Google search and we're going to do a Google advanced search. So have you ever heard a Google event search? [00:42:26] No. [00:42:27] Ok, so if you just go to the semantic search, I yeah, pretty much. Oh, so go to Google Advanced Search. So we'll start from the very basics, right. So we want to try to find keywords. Right. So right now, what are the words that you want? [00:42:43] You're looking for Data size projects and looking for in banking and finance. Right, exactly. [00:42:52] So we could put Data science project and then put it in quotes and then banking. And finance, right? [00:43:03] Um, this exact phrase, maybe we want to see a case study, right. And the reason why it's a case that is you want to see, are there any, um. Interesting use cases that we might not be aware of. Let's see what other people are doing. Let's see what Data scientists are doing in this space. What type of problems are working with any of these resources put in visualization? How about that? And let's see what results we come up with. So you get a number of results. Top it is how Data science in finance has increased the industry's profitability. [00:43:42] Right. [00:43:43] And kind of just read through a bunch of case studies, use cases like this so that you can understand, OK, how is how how's the banking industry using data science? And if you read enough, you'll start putting some dots together, start connecting some dots. Right. So that's step one. Try to figure out. What kind of problems are solving, right? Step two is going to be all right, great. Now I need Data, right? So we could type in. Maybe we can go to the World Bank. [00:44:19] Right. And open Data. And we know that the World Bank has opened Data. Right. [00:44:27] Well, a number of different types of data sets that you can dig into that, you know, that might not necessarily be banking finance related, like in private banking finance in that sense, but it's still Data still live Data you could still do interesting visualizations with it. Right. In fact, you might even you might even go to Data, the World Bank dog and just read through. A bunch of interesting use cases in and look for something yourself, like you see here, they've got to open dating a lot. They got Data Bank, they got the Micarelli Library, so on, so forth. So you can just look at the open data catalog and see what they have. And all of a sudden you're armed with a number of data sources. You're armed with an idea for what types of analysis are typically being performed by Data scientists with finance and banking type of data and modeling. The most important ingredient ingredient is like your own curiosity. Right? And you merge those things together and you can apply that through Tableau or through Python or whatever, just through a project and host that wherever you like, whether it's public Tableau or GitHub or whatever. But that's pretty much how you work through a a. Project idea type of thing, so it begins with just trying to read and understand what is going on in the industry of interest in. [00:45:55] And just finding Data and executing the grid. [00:46:02] One more question. [00:46:03] So if I want to start with something like the things that I already know I can work on and like you're saying, that do some research work and try to visualize and try to do things yourself and do some projects. [00:46:19] So if I ask you that, what are what should be the I don't want to do everything. I don't want to learn every single language or everything out there. [00:46:30] But if you go down to three languages, if I must say that I should do, what would you suggest? [00:46:40] Is this a roundabout way of asking me if you learn Python or. [00:46:45] Ah, the army, yeah, because I keep hearing all the time. [00:46:52] Yeah, explicitly said, don't ask me that question, but I'll answer it anyway. Just learn Python, just learn Python. I mean, why not? I'm saying that I'm sorry if you are listening in your Norrie's or if you're watching this on YouTube and you're an artist, like, I apologize. I just I like Python. But if you have time when I learned about this. Right, OK, I definitely use ah when I have to. It's not like I'm against it, but I'll use it when I have to. [00:47:21] But if you're just starting out I would say Python just because um. Because I personally like it. No other reason than that. Yeah. I do so much. Yeah. [00:47:37] Awesome. We got her. It is a recent. Data science graduate, currently interviewing at a company right on and they're heavily testing of case studies. OK, her question is, how do you solve cases during the interview where you have little or no domain knowledge about the cases come from a non-technical background with only Data science internship experience. All right. How how's it going to get. [00:48:06] Do you want to unroot? [00:48:09] Now, Harp, yeah, so the cases that I'm talking about is it's not even those opening questions, they're basically testing on the little Data said few columns. And what features would you like to experiment with and what kind of feature Glendinning that you would like to solve? So I've already given to dozens of interviews, but I felt that I could have definitely done better. It's just that within that 20, 15, 20 minutes time frame of time, I'm not able to come up with some personalized features or know think out of the box and come up with features. So so far, I've been lucky enough to get questions that are going to have some knowledge about. But, you know, there's a lot of, I would say, problems out there, like recommender systems, then communication problems, different kinds of classification problems. So how do I prepare in like a few days off, you know, in the next few days for the next individual round? [00:49:07] Yeah. So again, the job posting itself like a syllabus I looked through there and see exactly what it is that they're like specifying in that job description. Right. So that's step one. Make sure you totally treat the job description, job posting like it is a syllabus. Step two is maybe you can look up if you have the names of the data scientist that you'll be interviewing with, maybe you can look them up on LinkedIn. Take a look at their about section, take a look at what they have written as their job description on LinkedIn. Is there any indication that they're working with a particular type of tool or particular type toolbar of projects? Right. Because some people will put it straight up on their LinkedIn the different projects they've worked on. [00:49:57] Right. [00:49:57] And so that could be an indication as to what's in their wheelhouse, what they're comfortable asking. [00:50:06] Third is doing something very similar to what we just did with Gunjan here, which we can do again right now, actually, so just for illustrative purposes. I'm going to pull up LinkedIn again. So what is the industry that that last interview that you wanted to tell me, the company or just the industry or whatever you want to share? [00:50:29] So I think they are wanting to it's a startup. So they are working on providing these products to companies like what the product was, where they tried to recommend what and what a particular customer is going to order based on their past or those using facial recognition. So these are the kind of products that they launch for different companies that are MMD. [00:50:56] That sounds super interesting. They're using facial recognition to recommend products that. [00:51:04] Also then also their past, Ordos and Data, organized opposition to the facial recognition. I believe this is the super interesting. This is the product of the launch for Nestlé. So even I was quite intrigued by that. Yeah, that's really cool. [00:51:22] All right, so. [00:51:24] You said the interview is coming up, pretty had it, so I already gave it to you, don't I? Next don't have to. It will be another round of case study and I really need to stand out in order to correct this one. [00:51:36] So let's just deconstruct this a little more. So what did they did they give you any indication on what you might be? Covering in this last interview, like they explicitly say, we're going to do a case study. [00:51:50] Yes, something like that. So go on. Go on, tell me. Yeah. So it's going to be Data Sanski study last one. Well, they mentioned it's going to be a regulation and classification. So it was more of a statistical question that I was tested on. So I analyzed my answers and I felt that it was generic. I would suggest a solution that I had given, like I'm going to perform. I'm going to see the distribution of my baby to see if there are outliers, perform any logarithmic transformation, then check for missing. I'm going to be witnessing values. So I feel that these are the these were not is something that I would have any problem. [00:52:31] So, yeah, OK. And just to clarify, so this next question, the next case that you're expecting to build out a model, they're going to give you raw data and your end result should be a model. [00:52:46] So basically, they will be doing a set of Scholem's some knowledge about the columns. This is a data that you have. What features would you like best would be performing statistical modeling and then questions like how your data model biplane will look like. OK, what what algorithms would your best with this with this kind of data center? Even coming up with an algorithm like I can just see, you know, I'm going to test every algorithm in democracy, probably decision do might work better for that kind of problem. So how do you think in that? [00:53:18] But it was me I would tell them I'm going to use it and actually test every single algorithm before you could say that. [00:53:28] Yeah, yeah. [00:53:29] Uh, but but yeah, even I did somewhat similar to that. I will probably test with simpler algorithms, accuracy and then the complex algorithms at eight 'Ates as it's always better to try something simpler, which is much more explainable than jumping onto complex algorithms. [00:53:46] Yeah. So. Piling up on Google, so you do feature engineering recommendation systems, so you sound like you have like a hybrid recommender system, whether they're going to ask you that specifically or not, I'm not sure. But, um, that's a huge question. So let me just let me refer you to something else instead. Um, have you read the book Feature Engineering for Machine Learning? [00:54:12] No. [00:54:13] Ok, I think they actually let's see if I could find you. [00:54:23] I think it's freely available online. [00:54:27] Might be, yeah, OK, so hopefully this is it. Yeah, so here's the link, um. So. There you go. So there's a link to it and it's a great book, I would just take a look at this and, uh. [00:54:53] Oh, and they have built a recommender academic paper recommender, so that might be interesting for you to check out. Um, I guess without knowing too much about the industry, not knowing too much about. [00:55:14] What's going on, like it would just be in general, just understanding. [00:55:20] How to perform feature engineering would be useful. So definitely take a look through that book, because at the end of the day, like, you know, your future engineering is probably the biggest thing you could do to come up with a model that performs better than. [00:55:35] Having to make it more complex that I mean, oh, definitely read into that book and. [00:55:42] Pick up some of those techniques, maybe there might be stuff in there that you didn't know and try to find opportunities to apply that in in your own. [00:55:52] And it looks like somebody here posted a thing on LinkedIn. I don't know what this is, but we can pull it up. Uh, yes. [00:56:00] One of Eric Webers posts must be the post where he was talking about. [00:56:07] Uh. Yes, yes, exactly. This was the one. So Eric Weber has a post. [00:56:17] 50. [00:56:19] Titles that you can look for that do not mention Data scientist and you'll be doing Data science type of jobs. [00:56:26] So there you go. Cool. All right, everybody. Well, thank you so much for hanging out for office hours. I appreciate you guys coming by and spending an hour with me. If you guys liked office hours, were more than happy to have you as part of Data South Stream job. Here's the link right here. Extended for the weekend through the weekend, you could register for Data his dream job at 70 percent off. That's code, for example, artists 70. Not only would you have access to me multiple times a week, but we've got John Sebastian, we got Deonna and we got Chris and of course, Mr. Kyle and being very vibrant community. So check that out. Also, be sure to check out the @TheArtistsOfDataScience podcast. A lot of cool stuff on that. If you're listening to this on the podcast, then you're already checking it out in that case. Thank you very much. Got a lot of interesting stuff happening in the next few months. [00:57:33] Going to be some really cool people coming on the show. [00:57:37] So definitely stick around. We'll do this again next week on the 25th and looking forward to see there, this is going to be up on YouTube and up on the podcast itself, most likely on Sunday. So keep an eye out. Also, keep in mind that registering for office hours also registers you for my email list and newsletter, which will be coming up soon. So I'll be sending you guys a lot of cool insights, cool just topics and learnings that I think will enrich your lives. [00:58:14] So, again, thank you for being here. I appreciate you guys coming on and hanging out. Take care. Have a good weekend, everybody. [00:58:24] But.