Welcome to episode #181 of the Super Data Science Podcast. Here we go!
Today, we have Tim Lafferty from Velocity Group. Tim and I discussed about Data Visualization, Tips for Data Scientists Before a Presentation, and the Top Ten Things a Degree Won’t Teach You. Tim promises that after this episode, you’ll be less intimidated and less overwhelmed to get into data science.
About Tim Lafferty
Tim Lafferty is the Co-Founder and a Managing Partner of the Velocity Group. He is currently training companies to leverage data visualization tools to incorporate on their day-to-day operations. Velocity Group provides business solutions to improve the systems, communication, and efficiency among others.
Tim says that the above question should be the easiest answer that your data visualization process could answer. The goal: your target audience should be able to get a takeaway to create actionable plans after a presentation.
Data visualization is as valuable as the other parts of the Data Life Cycle. For a start, your clients will not look at your notebook where you keep your data. it’s rare that they’ll ask the intricacies of the process. Visualization is which what leads to the big questions. It’s a huge help in the problem-solving process of industries.
Going through the visualization step can also help you when you’re getting lost or side-tracked on working with the data. Sometimes, it might be difficult to get the entire picture of what you should discover in the end. Visualization could be the facilitator of data mining. You’ll be able to check if you’re doing it right.
Tim’s main visualization tool is Tableau since he believes it’s the most intuitive. He says that his goal is always to be platform agnostic and this is what Tableau offers. Principles should remain the same for all platforms.
Tim and I also give tips on what to do before, during, and after a presentation. Here are what we discussed:
- Use the 5-second Rule. If it takes more than 5 seconds for the audience to get what you’re showing them, you failed.
- Use complementing colors. Look into color
- Have zero text on the presentation. Or, if necessary, just one word to draw attention.
- You’re the focus. The dashboard is just there to assist you. The slides are just there to assist you.
- Choose appropriate graphic organizers. For example, you can’t always choose pie charts!
What if there are clients that insist on what they want despite the results? Tim introduces the Rule of Three. Firmly state your opinion twice. Make sure that they understand alternative routes. But, remember: in the end respect client’s decision and help them overcome future complications.
Lastly, here’s what you shouldn’t miss on this episode: Tim shares his own Top Ten Things A Degree Won’t Teach You. Make sure to listen as Tim explains it point by point!
GET YOUR FREE INFOGRAPHIC HERE!
In this episode you will learn:
- 3 Main Aspects of the Life Cycle of a Successful Data Project: (04:50)
- Data Architecture – modeling, proper structure
- Predictive Layer – ‘nerdy stuff’ (machine learning, algorithm, etc.)
- Presentation Layer – for your target audience
- How did Velocity Group start? (06:15)
- Visualization is the vehicle in which you transport your results to the client. (09:20)
- The power of storytelling in data visualization. (10:44)
- Every concept should be presented in a platform-agnostic view. (15:08)
- Visualization could help in making sure that everything on the back end is scalable. (17:08)
- Data Science Venn Diagram (23:00)
- Kirill’s and Tim’s Tips to Improve Presentation. (28:34)
- The Rule of Three. (39:12)
- Ten Things A Degree Won’t Teach You.
- It’s a collaboration, not a competition. (24:40)
- Occam’s Razor. (37:30)
- The ability to find the answer is more valuable than the answer itself. (41:46)
- Concepts over syntax. (43:30)
- Less is often more. (44:27)
- Common Sense. Does it pass the sniff test? (46:57)
- Take care of yourself. A healthy balance is critical in this industry. (49:07)
- Help and be helped. (54:50)
- 90% of your work will go unnoticed. (57:25)
- Staying stimulated and the importance of having fun. (59:38)
Items mentioned in this podcast:
- Data Science Venn Diagram
- SDS 176: The Importance of Storytelling in Data Science
- Confident Data Skills by Kirill Eremenko
- Machine Learning: The Art and Science of Algorithms that Make Sense of Data by Peter Flach
- The Kind Worth Killing: A Novel by Peter Swanson
- The Big Book of Dashboards by Andy Cotgreave, Jeffrey Shaffer, and Steve Wexler
Kirill Eremenko: This is episode number 181 with founder and managing partner at velocitygroup.io, Tim Lafferty.
Kirill Eremenko: Welcome to the Super Data Science Podcast. My name is Kirill Eremenko, data science coach and lifestyle entrepreneur. And each week we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today, and now let's make the complex simple.
Welcome back to the Super Data Science Podcast, ladies and gentlemen. Very excited to have you on the show today, and we've got a special guest, Tim Lafferty, who is the founder and managing partner at velocitygroup.io. Velocity Group is a consulting firm that's worked with some big clients such as Delloitte and Delta, Pluralsight's, and many more. Their website is velocitygroup.io. And what we talked about today is a lot of experience, or a lot of insights that Tim got from his experience in this world of consulting in the space of data science.
So, specifically, you'll learn about data architecture, the predictive layer, and the presentation layer, and what the differences are. Which one of those three is the most important, according to Tim, and why. We'll also talk about Tim's personal journey through data science, how he got to where he is, how he started his consulting business, and what they're doing these days. And also, actually, in this podcast, Tim prepared for this podcast, Tim prepared 10 things that a degree won't teach you. 10 tips for your data science career, which we went through one-by-one in detail, and I personally enjoyed a ton. In fact, we had so much fun going through these tips that we decided to put together an infographic.
So I literally just jumped off the call with Tim just now, so the infographic, as of right now, is still an idea. But when you're listening to this podcast it's already going to be ready, and you'll be able to download it. So if you're in front of your computer right now, then head on over to www.superdatascience.com/181 where you will find the show notes for this episode, and the infographic is there, so you can follow along with the episode.
If, at the same time you're listening in the car or while you're jogging, that's totally fine as well, because you can just listen to the episode, and then when you get home, remember to go to the show notes and download the infographic from there.
And, of course, there's going to be lots of fun and lots of laughs in this episode. Really enjoyed chatting to Tim and can't wait for you to check out all the insights that he had to share. And without further ado, let's dive straight into it. I bring to you Tim Lafferty, founder and managing partner at Velocity Group.
Welcome ladies and gentlemen to the Super Data Science Podcast. Today I've got a very exciting guest with me on the show, Tim Lafferty from Atlanta. Tim, how are you doing today?
Tim Lafferty: I'm doing well. How about you Kirill?
Kirill Eremenko: Doing well. Very much so as well and man, so excited to have you on the show. We've been trying it just before podcast and first thing I probably want people to know it your goal for this podcast. I loved it when you just stated that. Could you just state that again please, for our listeners? What's your main goal for today's show?
Tim Lafferty: Yeah. Sure. My objective with speaking with you today is really to make sure that a lot of newcomers in the industry have a less intimidated approach to getting in here. Sometimes I believe newcomers can have this feeling of, "Oh my gosh, data science community is very daunting. It's overwhelming." My goal is really just to make it more of a fun approach.
Kirill Eremenko: Fantastic man. I love that, and I'm all in so let's do this, man. Let's break the barriers and destroy the myths. You ready for this?
Tim Lafferty: Yeah. Let's do it.
Kirill Eremenko: All right. So first thing, I'll probably officially read out your length and title 'cause I love it and then you can explain it a little bit. Tim Lafferty is a wrangler of data, architect of algorithms and the destroyer of pie charts. So cool man. Tell us a bit more about that. What do you do in life, man?
Tim Lafferty: Most of what I do professionally at least is data centric and in my opinion, there's really three main aspects that make a successful life cycle of a data project. The first one, and maybe not necessarily the serialized, but the first one being the data architecture, the modeling side of things. Having proper structure. And the second one would be the predictive layer. So this is where all the machine learning takes place, the algorithm developments and all of the fun stuff and as I joke around, it just the nerdy stuff that we all like.
And finally there's the presentation layer. This being wherever you deliver your results to the attended audience. Those are the three fundamentals, in my opinion, that create a really strong foundation to a successful data project.
Kirill Eremenko: Got you. Got you. That's really cool. We got data architecture, predictive layer and presentation layer that make this whole thing work together. Before we dive into each one of those and discuss them in a bit more detail, let's paint the picture for our audience. I'd like for our listeners to know that where you're actually getting all of these insights from. That you're not just read a book and you picked them out of thin air. You're actually a magic partner and creator of a consulting firm, if I may call it such, of the Wall Street Group Development. And the website for those interested and what like we seem to say in computers it's velocitygroup.io. And you guys are as far as I can see and from what we discussed are quite a successful consulting business being around for five years and you work with companies like Delloitte, Delta, Pluralsight's and others.
Quite some big names there. Tell us a bit about the whole business and how it all started, so that our listeners know where these insights are coming from.
Tim Lafferty: Okay. Really what we founded Velocity Group on a more of a data architect business model and then had the layer of predictive analytics. One thing that we realized pretty early on is that a lot of companies are doing this. Machine learning is hot right now, man it's fantastic, as is data architecture in general. We weren't really having the traction that we wanted and all the sudden, something clicked. Actually a good friend of mine, Ryan Nokes with Lightpost Analytics asked me, "Hey are you interested in teaching Tableau?
This is a couple years ago. I decided sure. Why not? I had played with Tableau and it was an area that I absolutely needed some help on. I learned it with his curriculum, ended up becoming very comfortable with it. After I did the training I came back to the company, to Velocity and realized that's really the piece that we were missing is adding this visualization. And I think as backwards as it may sound, the visualization piece actually comes first in most of our engagements with clients.
For instance, a typical engagement might be that we go to a client, and we say, "How are your courts?" How are your BI visualizations? Touched some of those up and those lead to bigger questions like, "How can we fix this or what would happen if we changed this?" And then we can start right back at the beginning of data structures to predictions to visualization.
Kirill Eremenko: Got you. Wow. That's really cool.
Tim Lafferty: That's a fancy way of saying we stumbled into it.
Kirill Eremenko: Wow. Okay. That's a very interesting thing because I've actually noticed this on the podcast more and more that visualization is starting to play a higher role in the world. You started the [inaudible 00:08:57] company in the predictive layer then added data architecture shortly after, and but then the presentation layer is actually what made the big difference. Why do you think that is? Why do you think more and more so in today's day and age, the presentation layer is the actual missing component, the thing that makes or breaks a project or makes or breaks a relationship with a client?
Tim Lafferty: I believe that the visualization layer is the simplest. I don't mean that, in an offensive way, I just think that it's perhaps viewed as the simplest portion of the life cycle. But the truth is, I think it's one of the most important because it's the vehicle of which you transport your results to the client. And your clients don't want to look at your Jupiter notebooks. Right? Most of them if you gave them script like that, they're not going to know what to do with it. We as humans are just more visually inclined, so if you show them just stunning visualizations about their data they instantly have an emotional connection. And it gets them excited about doing more stuff.
It's pretty rare that I run into clients that really, really care about the back end technical stuff. The line by line code. Right? Of course to talk about bottles and what we're doing in terms of structure but I think, sometimes that's what they use us for. Right? They hire us in data science in general to take care of that.
Kirill Eremenko: Yeah. Yeah. I totally agree and actually too your point about clients not wanting just to [inaudible 00:10:46] Jupiter notebooks or code, I was watching a talk just recently by Ann Marie, I think. It wasn't her [inaudible 00:10:57], it was something else. This is a lady that talks about the power of storytelling. Ann Marie Houghtailing. That's her name. In one of her talks, I watch this little show, a short ... a snippet, she said that when you are just conveying information to someone especially in consulting, especially in technical discipline, there's only two parts of the brain of the person you're talking to that are activated. Don't really remember what they are called, they have some scientific names.
That's the way they're processing, recording information. But when all the sudden, you add visualizations, you turn that whole technical spill or your report that you're delivering, you turn into a story, then five parts of their brain are activated. And they remember and understand things way better. Have you noticed that since the time that you added the presentation layer? Have you noticed a better ... I would understand of course, your clients would be more enthusiastic about the results. But is the retention better? Do you have better insights? And do they get better insights out of it and do they understand the results faster than it was before?
Tim Lafferty: Yes. Yes. Exponentially. A good exercise for this is if you read out ten different statistics, or ten different metrics if you will, KPI's of a company. And we often do this in the beginning phase. And ask them to repeat them back to you now or later. Right? Go on, move on, talk about ... Very, very rarely can they remember more than three or four. However, if you provide a chart that's engaging, most of them can remember seven to eight. That's just in my personal experience. That's part of something we do in the sales process. Why should we do visualization? Right? We're using tables right now and they're working fine.
Well because you retain more therefore you can act on more. And you can turn your company and make it more of a proactive, data driven company that makes decisions based on seven or eight of those questions that I asked earlier as opposed to three or four. You're getting a better idea of the bigger picture.
Kirill Eremenko: Fantastic. I love that example. It's a live example of actually that concept people will retain information better if you show them the chart.
Tim Lafferty: Absolutely.
Kirill Eremenko: All right. Speaking of visualization, why Tableau? There's quite a lot of visualization tools out there. I personally am a huge fan of Tableau and we have several courses on the tool and I love the community, around it, But what's your view? Why did you settle ... I'm not sure if you're ... Is just Tableau? Is other tools as well? But like something-
Tim Lafferty: There's other tools. Tableau is definitely my main visualization tool and out of all the more self serving tools. Right? We're not talking about Python or R, or any of the charting libraries, but other than the stand alone visualization platforms, Tableau to me is the most intuitive. And the reason I don't necessarily do it for myself, but I do it for my clients. And I'm not paid to say that. But Tableau is easier for a lot of people to grasp because it just really makes visualization simple.
Kirill Eremenko: Got you.
Tim Lafferty: It just started ... Actually how I got in to Tableau right out of college, one of the ... The company I was working for, the analyst came in my office and said, "Hey I don't have ... I'm not able to go to this conference because of [inaudible 00:14:50]. I'm not able to go to this conference. Are you interested? I'm like, "Okay, Sure." And it was Tableau and from there it was all down hill.
Kirill Eremenko: Yeah. Tableau conference. I haven't been, but I'm really looking forward to going one day. They do an amazing job as well there.
Tim Lafferty: And I don't want to dismiss any of the other platforms either. My goal a lot of times in any sort of teaching or speaking engagements or really just best practices is to create a platform agnostic view. But of these concepts should be platform agnostic. Right? Whether you're going to click Power BI or Cognos or whatever, the principals should remain the same.
Kirill Eremenko: Yeah. Yeah. Totally agree. And it's very interesting and inspiring to hear your passion for visualization because rarely do we have ... Usually it's people either going to machine learning and AI and Data Science and that direction or they go into [inaudible 00:15:51] side of things. And that's fair. Both parts are great and that's what I've been trying to convey in the podcast as well. But here it's exciting because you, or a person who can do all the difficult to technical machine learning models and build all of that and even the architecture side of things, but at the same time, you acknowledge the power of visualization. So it's a message to everybody out there that's being focused a lot on the machine learning side of things and the deep learning side of things and the technical aspects and maybe the statistical side of things. It's a message that adding visualization to your arsenal of tools as a data scientist is not just going to be like a one plus one equals two equation, it's going to be a one plus one equals one hundred equation, because-
Tim Lafferty: Correct.
Kirill Eremenko: Exactly. There's a lot of synergies. Right? All the sudden you can take those same results that you had from machine learning exercises and present them in an amazing, beautiful way with powerful visuals and then tell that same story that you're about to tell but have that visual anchor for people to look at.
Tim Lafferty: Additionally, I think becoming a stronger visualist helped me in my more technical roles as well. Because I'm sitting here putting a lot of it off on the clients but the truth is I learn a lot when I look at visualizations. If I'm dealing with a dataset, sometimes it's difficult to get the entire picture. You have to go through multiple literations and as soon as I start putting it into visuals, I get new ideas and it sparks something. It helps all parties involved.
Kirill Eremenko: Exactly and I like what you said at the start as well that a lot of the time your engagement start with visualization. When I worked back in Delloitte, that often was the case because the visualization allows you to not only present the results at the end or create a dashboard, like an interactive tool for a client, but also at the start of the engagement visualization is a facilitator of data mining. Right? Like it's much-
Tim Lafferty: Absolutely.
Kirill Eremenko: Much harder to mine data when you're coding and pythoning, to check something, then recode it, check something recode it, or you'll looking at a table with a million rows and columns. It's much easier when you see what's the mission of Tableau to help people see their data. I think. So when you see your data all of sudden and you can interact with it then that's when the ideas starts popping up.
Tim Lafferty: Yeah. And a lot of these processes in my opinion should work in parallel. As I'm trading models and developing in Python, I might look at the results in Tableau, and then go back to Python to tweak it or maybe even adjust some of the structures that go in to the database. It's not a linear process.
Kirill Eremenko: Yep. Yeah. Totally agree. Okay. That's very exciting. How about ... Tell us how you got into data science in the first place. We know about your business and how that started and ... Well a little bit how that started and what you're doing now, but very interesting I always love to talk about the journey of a person. What did you study? Where did you come from and that professional sense and how did you get into data science?
Tim Lafferty: Well I'll say, me getting into data science was unintentional or it wasn't conscious at least. I was in Georgia Tech, and I graduated with a degree in industrial engineering and the concentration of the statistics. This effectively gave me a lot of time to focus on optimization and efficiency. Right? And so-
Kirill Eremenko: So sorry. Just to understand. Industrial engineering is like building buildings? That type of thing?
Tim Lafferty: No. I noticed that's funny. That's the common perception.
Kirill Eremenko: That's similar. Right?
Tim Lafferty: Yes. This is industrial systems engineering. The process of making things more efficient.
Kirill Eremenko: Oh. Okay.
Tim Lafferty: So whether it's optimization simulation or a lot of the [inaudible 00:20:01] algebra in there. It's a fantastic discipline that I think might be overlooked.
Kirill Eremenko: Got you. Okay. Good to know. Thank you.
Tim Lafferty: I graduated from there and I got a job at a large Internet auction site. They needed a data analyst. And that's a pretty umbrella term. Right? I already had the statistic background coming in and I was so fortunate to have a fantastic mentor of the DBA and he really got me excited into the scale ability aspect of things. Right? The modeling, performancing, there is where the birth of my data modeling love came from. All the sudden, I have all these fun algorithms going, and I'm able to, I'm pretty efficient in all sorts of data bases. But now what? So I can do that, that's great. And what I kept missing was the fact that there was no visibility in what I was doing. Or minimal. Right? People took me at my word which is dangerous.
I ended up getting into Tableau and from there I realized as you put those three together, that's ... I'm like, "Wait a minute, this is what data science is." Right? Making sure that everything on the back end is scalable. We have good performance, it's structured properly. I have a really solid models that I'm proud of and that are robust, and I'm effectively communicating this with the management team. That's a long way of saying that I didn't intentionally get into data science, just one day I woke up and, "Oh yeah, that's what I do."
Kirill Eremenko: Yeah. Yeah. Got you. Got you. That's really cool man, and exciting. I love how sometimes these stories are like you get into data science right out of the blue. You weren't intending on a career in analytics or whatever it was called back in day, business intelligence. And then just you met somebody like your mentor and he got you a little bit into algorithms and you added your skills together and there you are. You're in data science. That's really cool. Really cool how you got here.
Tim Lafferty: Well I think the term is again, it's just a blanket term. But that's a good thing. There's a lot of data scientist that focus more in one area than others. There's some that are evenly distributed and it's just a fantastic industry to be in especially at a time like this.
Kirill Eremenko: Yeah. Yeah. Man it's your story is very in line as you pointed out with your current philosophy of data architecture, predictive layer, presentation layer. And while you were speaking, I was thinking of that ... Remember the Venn diagram for data scientist? What does it have it there? It's got programming, statistics and mathematics. Is that ...
Tim Lafferty: Yeah. One of the best quotes ... Shoot I'm going to butcher this, but I was reading something, a definition. What is data scientist? And it was something along the lines of a statustiction, or somebody who can develop better than a statustiction, but knows more statistics than a developer. Right? It's just ... Yeah I'm sure I butchered it.
Kirill Eremenko: I think I know that one. It's a developer who ... Sorry, sorry. You're right it's hard.
Tim Lafferty: I know.
Kirill Eremenko: It's the statistician who can develop but or a developer who can do statistics?
Tim Lafferty: So see Kirill, this is exactly why we need a chart to show this. People would understand it better.
Kirill Eremenko: Oh yeah. True. True. True. Okay I found the Venn diagram. It's hiking, skills, maths and statistics and substantive expertise, domain knowledge.
Tim Lafferty: Sure.
Kirill Eremenko: I'm guessing ... As far as I remember. But yeah. Coding starts in domain. That's [inaudible 00:24:11] like commonly believed to be there. When you was speaking, I was thinking you should create your own Venn diagram and write a article about that scientist is actually on intersection on data architecture, predictive layer and presentation layer. Have a fresh look on this whole thing.
Tim Lafferty: Yeah. I think I'll do that. That's a good idea. You may bring up an interesting point. I like what you said about the main knowledge. A lot of times, I think that's overlooked some. One of the common scenarios that I run into is as consultant, you go into a company and perhaps people aren't as technical as you. And that's okay. Right? Or else I'd be unemployed. But I think it's very important to never dismiss other people just because they're not as technical as you. There's a lot of these things ... A lot of these engagements to where we can build out great models and have great results and great visualizations but I still need domain knowledge that I don't have from the person that's been there for 20 years.
I think it's very, very important not to dismiss domain knowledge or write somebody off, as perhaps not as valuable as yourself just because they don't have the skillset you do. At the end of the day, this industry is a collaboration. It's not a competition. Or that we can work with people really all across the space, I think the better the results we'll have.
Kirill Eremenko: Absolutely. I love it that quote it's a collaboration, not a competition. I think the data science community and I've had a few guest that have mentioned this and have talked about it, but data science community is one of the most friendliest, and most helpful communities out there, whether it's Tableau or whether it's in the Github or wherever you go. Even these in the sub pocket of business, every time you're welcomed, your people help you on some questions, and help future projects and I would love for us all to keep it that way. As you see let's keep it at collaboration, I think everybody can benefit by helping each other out, learning together.
Tim Lafferty: Well I think we do a good job and I really think that's what you do a fantastic job of is really having these complex ideas and making them user friendly to anybody who can [inaudible 00:26:37]. And I think that's really encouraging because it's easy to get lost in some of these technical details. Right? And that's one thing that I really try to be strong at, is thinking with a level of complexity while speaking with a level of simplicity.
Kirill Eremenko: That's good.
Tim Lafferty: This comes down to ... I remember my grandmother was telling me one time that she had gone to a few doctors for some illnesses and they were speaking so just in such a high level term. She's not a doctor. And I remember her being frustrated and to the point where she even said. I don't even want to go back to the doctor anymore because I don't understand them. And that's hard to hear and it makes you wonder how many businesses could have possibly been turned away because they're discouraged. They don't understand some of the data science lingo that we use. It just really opened my eyes to how important simple communication is. And simple don't mean dumb by any means. Right?
Kirill Eremenko: Yeah. Yeah. Totally agree. Man, it's sad, sometimes, when you have some great results and because they're not communicated well, they're just go and disappear. They don't, they don't-
Tim Lafferty: You don't ever want to scare somebody away. You don't want to intimidate them and I think unfortunately, there's perhaps some people think that the more, the bigger words that you use is the same thing as equating to your intelligence. And that's simply not true. But to a point I think we do well as a community to be welcoming to be not intimidating of them.
Kirill Eremenko: What's your biggest tip that you can share that you use in presentation? I'm assuming you do a ton of presentations for clients and training session and so on. When you have these slides on the wall, is it something about the slides that is the most important rule of thumb that you follow or is it during your speech or is it during the words you use, the terms you say? Is there one tip that you can share with our listeners that they can take away and apply in their own presentations?
Tim Lafferty: Yeah. I think the largest, the biggest tip that I have is use the five second rule. Right? If I show you a dashboard or a chart and it takes you more than five seconds to understand one point on there, I've failed. I've failed. It's that simple.
Kirill Eremenko: You mean not understand the whole dashboard but one point? A point that you showed out, pointed out ? Or?
Tim Lafferty: Yeah. If there's not a single point that ... And let's say that I have a dashboard with four charts.
Kirill Eremenko: Yeah.
Tim Lafferty: Okay. One point on there, out of the four, or maybe it's a continuous point, whatever. You should have an insight within five seconds. I think one of the things that I see a lot of times, is you have all these really cool insights and so you put 10 charts onto one dashboard. Right? And it goes back to just because you can, doesn't mean you should. I'd rather see two charts across five dashboards as opposed to ten on the one. And that would be my one piece of advice to people. And the other one's, because I'm on a ...
Kirill Eremenko: Yeah. Yeah. Keep going. Keep going.
Tim Lafferty: But another one is colors. Colors are great. They really make things pop. But just as much as colors stand out, uncolored to complement. Right? Uncolored things stand out too and what I mean by that is the triggered point so use your colors wisely. Typically, I like to have less than three to ... Maybe we'll say less than five colors on a dashboard.
Kirill Eremenko: Yeah. Makes sense. Makes sense and yeah they're tools out there like whereas the Tablaeu has this already built in that allow you to pick the color pallettes that actually complement each other. That work really well. There's another one, I forgot it now. Color lab or something like that. It's for data scientist to pick out colors. Anyway, a little may be included in the show notes if I remember. Okay. All right. I'll-
Tim Lafferty: Trust me. I have a designer. I'm not good at the color portion, so I have a designer on my team that helps me out.
Kirill Eremenko: That's smart. That's smart. Get some help there and for those out there who don't have a business and who are not building out a consulting enterprise like Tim here, taking on the world, you can always go ... For instance if you need help with colors, you can always go and ask the community and they'll help you out or if you are doing some freelancing work or things, you can outsource through tools like Fiverr for instance. On Fiverr can get designers to help you out of make coloring something in for $5.00 to $15.00. There's always ways to if you're not ... I'm not good with design. I can see a good design or some bad design, but a great one, not the best person. So yeah, you just go and find some help on that front.
Thanks for sharing those tips. I also wanted to add my tips on the topic.
Tim Lafferty: Yeah.
Kirill Eremenko: And mine would be that when I have slides on the ... When I'm presenting slides and it can be a chart or it can be like I'm presenting the whole presentation and maybe some slides are not charts but a word and a picture or something like that. My rule of thumb is let's have zero text on my slides. Or almost ... I might have one word just to draw people's attention. I never put on very rarely zero [inaudible 00:32:52] would I put a sentence if maybe it's a quote by someone that I want to read out. Because as soon as you put more than a sentence on a slide and you expect people to listen to you, guess what they are doing? Their brains are reading the slide that just appeared and you're not talking as fast as they're reading and they're this whole misdisconnect and by the time they finish reading, you're still half way through what they already know, and then they get disconnected and they switch off.
And that ties in with my philosophy that when you presenting, for instance, especially visualization might be tricky because in [inaudible 00:33:28], people might think, "Okay. Great this dashboard that's my deliverable. No that's incorrect. Your deliverable is the insights that you're about to tell people. Your dashboard is there to assist you. My philosophy is you're not there to assist your slides. Your slides are there to assist you. All the time while you, whether if it's on stage of in front of the boardroom meeting or just in the meeting room, presenting to people, all the time the focus should be on you and then your slides should be like an axillary thing that you switch aside. People look at it quickly, they get the picture and they'll listening to you, they're focusing on you. In cases when ... Some people might ask, but what if I'm delivering an interactive dashboard the client needs to use. We'll that's great.
In that case, you need to create this dashboard and have it as a deliverable. But for your presentation, you need to change things. You need to either take screenshots and put them into position, or you need to know exactly the sequence of steps that you're going to go through dashboards because when you deliver a dashboard, indeed, the dashboard is the focus of the people when they're going to be using it and reading it. But when you delivering a presentation, once again, you have to be the focus, otherwise you're going to lose that presentation, and you're not going to get the insights across.
Tim Lafferty: I think that's fantastic. You couldn't have said that better. That's a great point.
Kirill Eremenko: Thank you. Thank you. And hopefully people combine our insights that we just shared then we'll get some great results and produce-
Tim Lafferty: We'll get less pie charts.
Kirill Eremenko: Less pie charts. I've heard that quite a lot actually. What's the ... How strong is your hate for pie charts?
Tim Lafferty: It's not. I joke because I think it's a community joke. But the idea here of pie charts. Why would you use a pie chart as opposed to ... Let's just take a bar chart. The simple explanation is whenever you get a pie chart with multiple measures in it you're then actually comparing areas. Right? Based on ratings not absolute degrees. Because if you have something ... Let's take three lines. Right? One's at 50%, one's at let's say 22% and the other's at what? 28%?
Kirill Eremenko: Yep.
Tim Lafferty: That's a lot harder to detect without labels of course as opposed that you have them stacked side by side. I don't necessarily have this strong hate for them, I just think that there are better ways to convey percent to totals. And ... Or proportions. I think one time that I do enjoy using them is if there's only one measure. Right? How close are you to your goal. Okay. If you want to do a donut chart, I think that's fine, that's still [inaudible 00:36:22].
Kirill Eremenko: Yep. Got you. I totally agree with that. For instance, I try to stay away from pie charts if it's more than three components or as you say, if it's hard to detect. Sometimes I'll throw it in just because people are used to, especially when you present to a board meeting, people are used to them. If you have that splints between two things, sometimes it can be okay.
Tim Lafferty: People are resistant to change. Aren't they? Like you said, if they're used to pie charts, some of them still want pie charts. And to that same degree, it's why I still use, as long as it's in taste, I still throw text tables on it. 'Cause people have been conditioned they want their spreadsheets. There's a balance.
Kirill Eremenko: Got you. Totally agree with that. Let's talk about an interesting thing that you mentioned on your list of things that you might be interested in talking about. Occam's Razor. Tell us a bit more about that. It is such a cool name for a concept and I'd love to hear how you apply it in data science. Maybe let's start with what is Occam's Razor.
Tim Lafferty: The idea that sometimes the simplest things is the correct answer. There's no need to make it more complicated than it needs to be. Right? And a lot of this, for instance, I think one of the better examples is a few years ago, and I think it's died down now, but there's this huge surge of big data. Everybody wanted to prepare their environments for big data. They want infrastructure and all that, but it almost became a joke really because I recall a company wanting a new back-end built. And they said, "Oh yeah, we have a lot of data we need to throw all these complex stuff at it." I said, "Okay, well what kind of size are we talking about?" We have over two million rows.
Well I can't think of a single environment that's not really scalable in. There's no need to have any sort of ... They had heard buzz words. Right? To do and start things like this so they wanted to complicate things unnecessarily. And I think that's really what we're trying to convey here is sometimes the simple approach is absolutely acceptable. If not optimal.
Kirill Eremenko: Interesting. Is that hard to do for a client or for a consultant when your client wants something huge and that's a lot of work for you and ultimately as a consultant you need to find the work? But at the same time you realize that's not the best thing for a client and the ethical choice to do is to tell the client that, that's not what they want. But also it's not hard just because of that, but it's also hard to ... They might have already had a board meeting. That might have already decided on this. They have their trajector I know and suddenly here comes Tim and then like, "Bam, it's time to change the trajectory." How does that work?
Tim Lafferty: I have a rule of three. Because you're right at the end of the day, I'm running a business, and if somebody is ... They want to hire me to do this project a certain way, "You're the client." However, I will state my opinion strongly twice and then make sure that they understand perhaps the alternative routes and some of the possible complications in the future and then finally about the third time if they really want to pursue it, I'll make sure that it's in writing so that they fully understand that the people that they paid to give them good advice gave them what they thought was the best advice and they chose to go against it.
And that's fine too. Sometimes it might be a coin flip, but we're pretty firm to what we believe our best practice is.
Kirill Eremenko: Very true. And fair enough. I think it's a good approach because as a consultant as well, some people might be like, "No, this is the wrong thing, I'm not even going to do it." But at the end of the day, everybody can make a mistake. Everybody can be wrong. Right?
Tim Lafferty: Sure.
Kirill Eremenko: You and I can also be wrong even though probably given that you've seen so many of these engagements and so many of these clients, you know that this is not going to end well or something like that. But at the end of the day, there's maybe a 1% chance that they see or know something that we don't know yet. And because of that, that's why they're so set. They're just not explaining it well or something like that.
Tim Lafferty: There's an art and I'm not even close to have mastered this, but there's an art to the discovery process. Because exactly like you just said, I can get in your environment and poke around and see what I need to see, but it seems like there's just always something that creeps up last minute. So I haven't mastered that art but you're right. At the end of the day, they make the decision to go a certain route. I respect it and we'll overcome the obstacle that comes.
Kirill Eremenko: Got you. Okay. Next item on your list was the ability to find the answer is more valuable than the answer itself. And these are so cool. You should write a book of quotes or something like that.
Tim Lafferty: I feel very strongly about this idea. I joke because my family thinks that I'm just a computer guru, and I'm not. Anybody who knows me, "Oh my gosh, he's awful." Whenever they ask me to fix computers-
Kirill Eremenko: What are you doing in this business if you're awful with computers?
Tim Lafferty: Let me specify. Maybe hardware. Right? I'm not a hardware guy. They want me to fix their computer, I google it. And I'm not by far special in that field but I know how to google things. And I think most people in our position, I can do the same thing. We may not have the answer off the top of our head but we can figure it out. We have good resources. We have a strong network and we're eager to listen, we're eager to learn. And we know how to ask the right questions. And I also think that, that's a stronger and more desirable value as opposed to just inherently knowing something.
Kirill Eremenko: Yeah. I totally agree with that. From my personal experience, I'd say that, that's how I chose my bachelor's degree. I didn't' know what I wanted to do if life, but then I thought, "All right, how about I go and do the most complex thing I can find." Which was physics and I really enjoy physics. So it's like, "If I can learn how to do physics, then that will teach my brain how to do stuff." Right? "And then I can learn how to do anything else." As, you say it's much more valuable to be able to find out how to do 100 things than know how to do five.
Tim Lafferty: Just moving along with that same concept, I'm a firm believer in concepts over syntax. Right? When I was in college, I spent a lot of time learning R and becoming fluent in all of that and the truth is the less that I used it, or maybe I didn't use it as much, I started to lose things and then I decided to go ahead and become proficient in Python but even to this day, as I focus more on visuals, I've forgotten a lot of the syntax and it's just not. I can go back and google. Right? I understand the concepts but I wasted a lot of time learning every small piece of framework language there.
Kirill Eremenko: Got you. Exactly. That is a wonderful one. Okay. Next one on your list is, less is often more. I think we talked about this. This is about being effective with visualization. Correct?
Tim Lafferty: Yeah. Absolutely and it's ... Back to the point that you have 10 different insights that you want to convey, perhaps you do that across multiple dashboards or stories or whatever you're deliverable is. And just in general with visualization, the simplest question to ask is, "So what?" So what? A pretty visualization means nothing in business context if there's take away. If there's no actionable item on there. If people can't leave your presentation with a take away, more so if they're excited to do it, then in my opinion that's not an effective visualization.
Kirill Eremenko: Yeah. I totally agree. And for those listening who may have not caught on yet, we're going through a list and I probably should have mentioned the name of this list at the start of this discussion so Tim put together a wonderful list called 10 things a degree won't teach you, and already gone through the first five. It's a collaboration, not a competition. We talked about it at the start. And then here now we've talked about Occum's Razor, the ability to find the answer is more valuable than the answer itself, concepts over syntax and less is often more. And we've got five more coming up. So I hope people are enjoying these and maybe writing them down if you're not driving. Don't write and drive.
But Tim I had a suggestion for you. In line with our what we were in the podcast so far that people remember things better in a visual. Might you have a visual already for these 10 things?
Tim Lafferty: I don't. No. But it's on the to do list.
Kirill Eremenko: How about we do that? How about we get together and we get some designers on board? Your designer, our designer and we make this happen. So when people listen to this podcast, they can go to the show notes and download it there and after the session they'll have a good solid take away with the 10 things that they've learned from this session that a degree won't teach them.
Tim Lafferty: I think that's a fantastic idea. I'll follow up with you afterwards.
Kirill Eremenko: Yeah. Sounds good. All right. Everybody listening? You'll be able to find these at the show notes and the show notes will mention ... I'll mention them at the intro, outro. You'll know which episode number to look for. All right. So we got five done. Let's do five more. You ready?
Tim Lafferty: Yep.
Kirill Eremenko: All right. Next one is common sense, does it past the sniff testing? I'm really excited to hear about this. What's the scent?
Tim Lafferty: The concept of the sniff test doesn't make sense. Right? Just stepping back to the average person. A lot of times what happens I find is that we get so involved with a project, we're so intimate with this particular data set that sometimes we forget to take a step back. You may have done everything properly, you may have done your analysis properly, your visualization are awesome, but at the very first stage, you could have arbitrarily done an outer join instead of an inner join, and now you have this 300 billion rogue dataset and not that your analysis is flawed at this point, but it doesn't pass the sniff test.
The example I like to give is that I was at a conference one time, and this gentleman speaking and mentioned something about active users on the website and he said that at any given moment, we have three billion active users on the website. And I'm going to go ahead and tell you it wasn't Google. Okay? There was a couple of looks around, anyway about a year later, I ran into a guy who worked with this gentleman. He said, "Oh yeah, he ended up having a wrong join in there somewhere and bloated his records. Right? His presentation was fantastic and some of his insights were great from a structural standpoint, but I don't think he ever took the time to step back and say, "Wait a minute. Does this actually make sense?" Because sometimes at the end of the day, the data that we're working in sometimes is our contextually ambiguous. Numbers are numbers. That's my way of saying just step back for a second and make sure that it passes the sniff test, the eyeball test, whatever the expression is.
Kirill Eremenko: Yeah. Got you. That's definitely a good, good piece of advice. Three billion users. Yeah. Definitely something went wrong there. Next one is take care of yourself. A healthy balance it critical in this industry. Very interesting and the unconventional wisdom. I think we all know that at the bottom of our hearts but not everybody points it out. Tell us a bit more about that.
Tim Lafferty: I love what I do. And I think most people that deal with data or in this industry, we love what we do. And it's very easy to become obsessed with the project and really just focus on it and give it 90% of your effort or 100% of your time. But the truth is that leads to brain fatigue and I'm a big fan of stepping back and taking care of yourself whether it's physically or emotionally. Whatever the case may be. I'm a avid mountain biker. Right? That's my way of escaping technology. It's important to me to actually get out away from the computer unplug and then a couple hours later, come back to the same problem. And that helps me overcome so many obstacles that I run in to.
I think everybody understands their moments and they know when they're just churning wheels. There's a point of dimension you turn. Right? Where you're just not efficient. It's more effective, it's better for you just to walk away for a couple of hours and make sure that you're mentally in the right place. And I mentioned the physical part, but there's other things too. Reading. I was really bad at ... I like to read books, but a lot of the books that I read were still technical. And so recently my wife told me she said, "Hey, start reading fiction." I did. And it's nice. It's nice to have that separation of the technical complexities with almost an escape. And I think it's very healthy to have that.
Kirill Eremenko: That's so cool, man. I've been the same. I hadn't read a fiction book for years if not more than a decade and then all the sudden, I was like, "Let's try it out. Let's see if I can get something out of this." And I think, earlier this year, I read ... What is it called? My first fiction book in ages. Andrew's Game, then some other book from [inaudible 00:51:31] and-
Tim Lafferty: Oh yeah. Those are great.
Kirill Eremenko: Yeah. Those are great. And then now I'm reading Assassin's Apprentice. I'm actually listening to them on audio book audible. I'm like, "They really suck you in and you completely forget about everything else." Especially if you're on a plane or in a car. You're driving or maybe somebody else is driving. And you for an hour or sometimes I'll listen to it for 40 minutes or an hour, you completely, your brain switches off in the sense of all this data science, technology business and the work stuff and things like that. I completely agree with you.
Tim Lafferty: It does. It creates a fantastic balance, and I'll tell you one more example. The other day, my wife and kids, we were in a craft store. Right? And just out of some random whim I decided to purchase a few canvases and some acrylic paint and some supplies. Right?
Kirill Eremenko: As you do.
Tim Lafferty: I ended up getting home and I started to paint and of course I'm getting questionable looks from my wife and kids. And I'll tell you, you'll never be surprised at what you're good at when you try. But painting was not it. I was awful. It was so terrible. But it was so funny therapeutic that it really helped. It turned off that side of the brain and allowed me to be more creative.
Kirill Eremenko: That's really cool.
Tim Lafferty: You're not going to buy my artwork anytime soon. It's terrible.
Kirill Eremenko: Oh, man I don't know. Sometimes they sell these artworks for like $30,000 and it looks like a five year old painted it. No offense to anybody. Maybe I just don't understand some contemporary art, but yeah. Have you seen Untouchable? That movie.
Tim Lafferty: Untouch? Which one is that?
Kirill Eremenko: Untouchable. It's a French movie about this man who got paralyzed in a parachuting accident and then he was looking for somebody to take care of him and then they end up bonding and-
Tim Lafferty: No. It sounds good though.
Kirill Eremenko: It's a good movie. It's a bit tragic, but it's a good movie. But anyway there ... 'Cause he's in the circles where ... The aristocratic circles of France and then go to business circles and he helps this guy that's taking care of him. He gets him to paint a painting and then he sells it for ten thousand Euros or something. It was his first painting.
Tim Lafferty: Well, all your listeners are going to go to velocitygroup.io, and it's going to be an art gallery pretty soon.
Kirill Eremenko: Yeah. You just found your new profession. That's cool. All right. And I wanted to ask you a personal ... How many kids do you have?
Tim Lafferty: Two. Ages two and three.
Kirill Eremenko: Oh wow. Congratulations.
Tim Lafferty: Thank you.
Kirill Eremenko: Best ages ever. You're probably getting so much sleep these days.
Tim Lafferty: Oh, they're fun.
Kirill Eremenko: Yeah. All right. And, which book are you reading right now in terms of fiction?
Tim Lafferty: Some Worth Killing. I believe is what it's called.
Kirill Eremenko: Interesting.
Tim Lafferty: Yeah. My wife was a little bit worried because the premise of the book is a man who is trying to find a way to murder his wife. I didn't choose it though. It was a best seller and somebody recommended it. But it's good to escape for half hour a night or something.
Kirill Eremenko: Got you. Okay let's get back to all this. We have three more. Number seven no eight. Oops. Eight. Help and be helped. That's a good one.
Tim Lafferty: Yeah. You hit this pretty early, but I think as a community, we do a fantastic job of just helping each other, and I think we need to make sure that we are also helping those who need help, but that we're also willing to accept help. Eventually a lot of people are going to either retire or worse become managers, and you need to make sure that you-
Kirill Eremenko: Become managers. I love that.
Tim Lafferty: ... bring the appropriate people. Right?
Kirill Eremenko: Yeah.
Tim Lafferty: I think we do a great job, but it's just important to be conscious of help when you can, and be willing to ask for help when you need it.
Kirill Eremenko: Got you. I totally agree with that.
Tim Lafferty: I think you do a fantastic job. That's one reason I've been following Super Data Science for a while and I always see how active you and your team are and I've never once detected any sort of condescension on your part. I think you guys do a fantastic job of just really giving back to the community as a sense of knowledge. I just want to thank you for that-
Kirill Eremenko: Thank you.
Tim Lafferty: ... and all your contributions. That's awesome.
Kirill Eremenko: Thank you. I really appreciate the comments. But I also love your ... In this point that you said, help and be helped. I like the second part, be helped. A lot of time it's people ... Even if you're helping a lot of others, it's important to recognize and not shy away from it when you realize that you need help with something. And [inaudible 00:56:27] beyond data science. It might be with work or with learning something in data science, but also might be in your personal life or in sports, in your diet and your health and whatever it is.
Tim Lafferty: Sure.
Kirill Eremenko: Yeah. A lot of people don't take ... It goes back to your whole thing, the previous point, take care of yourself, part of it is reaching out to ask people, "Hey, I got this problem, can you help me, help solve it."
Tim Lafferty: Well so let me ask you this personally. If somebody ever came to you or your team and said, "Hey I'm not really sure how to do something. Can you help me out?" Have you ever been upset that somebody asked you for help?
Kirill Eremenko: No, never. If I can, I always help.
Tim Lafferty: And most people are always excited to help.
Kirill Eremenko: Yeah.
Tim Lafferty: People enjoy helping and it's okay to accept help and equally ... I enjoy helping people, so I assume the inverse is true.
Kirill Eremenko: Yeah. Exactly. Okay number nine. And 90% of your work will go unnoticed. Interesting. Very, very intriguing. What do you mean by that?
Tim Lafferty: Well, don't quote that stat. It was more of a point.
Kirill Eremenko: A data science's stat. Okay.
Tim Lafferty: Yeah. There's not citation there. No, the principal there is I think a lot of people more on the visualization science, so let's take Power BI or Tablaeu developers. They think that perhaps they're going to go in and do nothing but visualization development for 40 hours a week and the truth is, a lot of times I think that you're spent in the back-end. Whether it's in [inaudible 00:58:04] or post ... Whatever your database is and building a lot of the back-end, we're optimizing things and a lot of that goes unnoticed. Right? And this circles back with my entire business model. I can write effective scrips and build effective models and have efficient databases, but at the end of the day, what people notice are my visualizations. And the other stuff is just, "Oh okay, they got it done." That's it. All of that hard work where I spend all of that time is ... Maybe unnoticed isn't the right word, but the simplest, least time consuming part, the presentation layer, that's what people notice.
Kirill Eremenko: Yeah. Definitely and you got to be prepared for that. Right? You got to be ... You got to know that, that's what you get into. That's just the nature of data science. You're going to be spending a lot of time on data prep and other things, what not and then just ... But if you do that right, the rest is actually fun.
Tim Lafferty: Oh agreed. Absolutely. Yeah.
Kirill Eremenko: The analytics and so on. But if you do it wrong, then you might have fun, but you'll end up with three billion active users [crosstalk 00:59:23].
Tim Lafferty: Well it just takes a while. You said it actually, prep in general. Right? So especially a lot of companies think that their data is already standardized and normalized. It's often not. I'll leave it at that.
Kirill Eremenko: Yeah. Totally. And final. Ten out of ten. Staying stimulated. The importance of having fun.
Tim Lafferty: Yeah. Back to just really being obsessed with projects getting too deep and too close, I think it's really important to still have fun with data. And I'll tell you something I did recently because I just wanted to practice something new, is going back to my mountain biking thing, I released a survey to a couple social media sites. Just a quick ten question mountain biking survey. And much to my surprise, it had massive trash. It was awesome. People all over the world, were answering this and it was really, really humbling. Honestly. But all the sudden, I have this real data set from real users, that now I can sit here and provide visualizations. I think that I recently made a post on at least the first half of the visual. And it recharged me because it's something that I enjoy. It's something that I can relate to. And it just made visualization fun again.
Sometimes data can be monotonous. I don't think that's a secret. It doesn't mean we don't love what we do, but it's having fun is the key to staying fresh and really energized.
Kirill Eremenko: Yeah. Well that's really cool. Even with your data science skills, you can have fun ... Not in a work but more ... I think it helps remember as well that what you're doing is not just for work, that it's actually a huge, a very powerful tool that you can apply in your own hobbies like in mountian biking or maybe to help somebody. Right? You could go and analyze ... There's these organizations, that allow you too to haggle but they allow you to analyze data sets for non for profits. And you obviously don't get paid or anything like that, but you actually help out an organization. It might be a school somewhere in Africa that's ... They're trying to analyze how many backpacks they're going to need in the coming semester or something like that.
Tim Lafferty: Yeah. There's some fantastic resources out there. I just think it's good to have some practice on some non corporate sense of data.
Kirill Eremenko: Exactly.
Tim Lafferty: Most of this stuff even if you're building models or whatever the case may be, you can apply those. Right? You might get new ideas to where you can take back to your day job. You're just more inclined to explore data that you're interested in.
Kirill Eremenko: Yep. Yep. Totally. Well okay. That brings us nicely to the end of ... Slowly to the end of the podcast. Those were ten amazing things. Thank you Tim so much for sharing. How do you come up with them?
Tim Lafferty: I learn as I go. And these are all things that I didn't realize as I came into the industry. And it's just about being aware of what's going on. I've been so fortunate to have just so many great people with such a network help me. And I think most of them agree with a lot of these points and that's where we got them.
Kirill Eremenko: That's awesome. And speaking of people and help and network, where would you say are the best places for our listeners to contact you or get in touch for your career? Or maybe they're businesses out there that need your help. Speaking of getting help then want the help of Velocity Group.
Tim Lafferty: Sure.
Kirill Eremenko: What's the best places to get in touch?
Tim Lafferty: You guys can always email me at [email protected] where is our website if you want to take a look is velocitygroup.i.o. And then follow me on LinkedIn. One of my goals this year is to be more active from a social media standpoint. I'm working on it, but you can find me at [email protected]
Kirill Eremenko: I'm just looking at you like "Tim, I noticed you know Andy Kriebel already. I was literally while we were talking and you mentioned Tableau, I made myself a note to introduce you to Andy Kriebel. He's a Tableau Zen master several times and works at the data school in the U.K. How do you guys know each other?
Tim Lafferty: I follow his work. I think he does fantastic stuff for the community. I don't know him personally, but I think he's just such a viable assets in terms of content and teaching. I think his style's great, I think his visuals are great. So I just really like following him and the things that he does.
Kirill Eremenko: Awesome. Well I'll introduce you guys. I think you guys have a lot of stuff in common with your passion for Tableau.
Tim Lafferty: That'd be great. Thank you. I'd appreciate it.
Kirill Eremenko: Okay. Awesome. And also of course, speaking of getting in touch with Tim, maybe Tim will be at Data Science. Go this year and you'll be able to meet him in person there. Right Tim?
Tim Lafferty: Yeah. I'd love to. I think that's the plan initially, we got to work out some logistic things, but yeah we have intensions of it.
Kirill Eremenko: That's awesome. That's awesome. And yeah to finish off one more question. What's a book that you can recommend to our listeners that will help them in their careers or just in life?
Tim Lafferty: The first thing that I'm going to say is I actually have your Confident Data Skills book.
Kirill Eremenko: Thank you. This is the first time it's been mentioned on the podcast. Thanks man.
Tim Lafferty: I don't want listeners to think that that was premeditated or something, but one thing that you did so fantastically in this book was really simplify I think, data science in general, especially when it comes down to the analysis section, I believe there's two parts to it? I think you did a great job of really partitioning and just breaking off the different types without going into details. I think you found a fantastic balance and all of my team has a copy that would quickly get the answers in need. I just wanted to say that's a fantastic book.
Kirill Eremenko: Thank you man. I'm so humbled and I'm so grateful that you took the time to read it. Thank you so much.
Tim Lafferty: It's an easy read too. Going to the total opposite, one of the more difficult technical reads that I have but I still think it's fantastic, is a book by Peter Flach or Flack, but it's called Machine Learning. The Art and Science of Algorithms that makes sense of data. This is pretty technical. A lot of it goes down to straight statistics, not even development or syntax. No code, more just conceptual math, but I think it's really, really, well written. You'll need a brain break every hour or so.
And then another one that I like is The Big Book of Dashboards. I like that for reference. And that's the one with the, I believe it's Steve Wexler, Jeremy Shaffer, and Andy Cotgreave.
Kirill Eremenko: Okay. Wow.
Tim Lafferty: That's a neat book that I reference whenever I have ... I need some inspiration for creativity. I think they do a great job of conveying effective messages without polluting the dashboard environments.
Kirill Eremenko: Well that's a good list. Thank you. And very diverse, so to recap. Confident Data Skills by yours truly, Machine Learning ... Who is that by?
Tim Lafferty: Peter Flach. Flach.
Kirill Eremenko: Peter Flach. And we might be pronouncing it wrong. I'm not sure about that. Flach. Yeah. And The Big Book of Dashboards. That's really cool. Okay. Thank you for sharing those as well and once again thank you so much for coming on the show. It's been an episode longer than usual, but I'm sure our listeners will agree, well worth it. I learned a ton and really appreciate you coming and sharing all these insights, Tim.
Tim Lafferty: Well I really appreciate you having me on, Kirill and I look forward to following you throughout your well established career.
Kirill Eremenko: All right. So there you have it. That was Tim Lafferty from velocitygroup.i.o. I really hope you enjoyed this podcast thoroughly and got a lot of insights. My personal favorite was the concept that the presentation there is the most important and it reflects the situation in data science that there is a lot demand, there's a lot of supply of skilled people who can [inaudible 01:08:40] but that gap between the insights and the decision makers is still there and that's why we need more and more data scientist who are actually skilled at conveying the insights to the end user to the person they're presenting to, in effective way.
And of course we've got the infographic. Make sure to head on over to superdatasciene.com/181 to download it there, so I'm sure by now by the time you listen to these episodes, our designers have prepared something super special, super unique for you so you can keep those ten tips close to your heart and keep them there. And remember as we discussed in this episode, people retain things better when they see visuals when they actually can relate to them so if you want to remember those ten tips better than download the infographic and maybe even send it to some of your friends, some of your colleagues who are in a space of data science and help them out. This goes in line with one of those actual ten tips that we discussed which is help and be helped. I think that was tip number eight. There we go. On that note, make sure also to check out velocitygroup.i.o if you're interested in any consulting engagements especially in Atlanta area or in the U.S.
And Tim mentioned that they will be expanding globally, but if you're in the U.S. especially in Atlanta, then make sure to connect with Tim and [inaudible 01:10:10] sessions and do lots of consulting very interesting consulting in [inaudible 01:10:13] and might be able to help you out. Also make sure to follow Tim on LinkedIn. It's Tim Lafferty and we'll have his LinkedIn URL in the show notes at superdatascience.com/181.
And Finally if you enjoyed this podcast, and if you got something valuable out of it, then if you know somebody else who can get something valuable out of it. If you know somebody who can benefit from these ten tips or from the [inaudible 01:10:44] of the presentation and why it's important or maybe who would benefit from connecting with Tim then make sure to forward them this episode and spread the word and help others also be successful data sciences. Remember it's a community where we're not competing but we're collaborating. So the more people you help the better it's going to be for everyone. Then eventually people will be helping out each other as well.
So on that note, I hope you enjoyed today's show and I look forward to seeing you back here next time. Until then, Happy analyzing.