Every week, we talk about important data and analytics topics with data science leaders from around the world on Facebook Live. You can subscribe to the DataTalk podcast on iTunes, Google Play, Stitcher, SoundCloud and Spotify.
This data science video and podcast series is part of Experian’s effort to help people understand how data-powered decisions can help organizations develop innovative solutions and drive more business.
To keep up with upcoming events, join our Data Science Community on Facebook or check out the archive of recent data science videos. To suggest future data science topics or guests, please contact Mike Delgado.
In this #DataTalk, we chatted with Eric Weber, Senior Data Scientist at LinkedIn, about why creativity makes or breaks you in data science.
Mike Delgado: Hello, and welcome to Experian’s weekly Data Talk a show featuring some of the smartest people working in data science. In most of our shows, we’re talking about the technical aspects of data, but today’s a little bit different. Today we’re talking about the art of data science. In fact, today’s topic was inspired by a blog post that Eric Weber wrote on LinkedIn. If you guys aren’t familiar with Eric Weber, he is the Senior Data Scientist at LinkedIn. He’s also an adviser to the Master of Science and Analytics Program at the University of Minnesota.
Just to give you a quick summary of his academic background, Eric got his Master of Science in business analytics from the University of Minnesota and then his Ph.D. in mathematics education from Arizona State University. I want to encourage you all to follow him on LinkedIn. The short URL to find him is ex.pn/ericweber. Eric, thank you so much for being our guest today.
Eric Weber: I’m happy to be here. I know we’re both on the West Coast, so 8 a.m. Maybe if people are watching this from the East Coast, happy late morning and then almost early afternoon.
Mike Delgado: That’s right. We found out before the chat that Eric and I are both early birds. We love getting into the office really early, so this is our peak time.
Eric Weber: Yeah. Everybody who’s watching is thinking, “What? 8 in the morning? Well, not me.”
Mike Delgado: Exactly. Eric, I thought it would be great if you could share with us your journey and what led you to start working in data science.
Eric Weber: I’m happy to do that. I think most people in this field, they would almost all say they have a nontraditional route because I don’t think there is a traditional one at this moment in time. I’ve always been passionate about education. I think that continues to be an undercurrent to everything I do. Along the way, undergrad, I focused on mathematics. I focused in grad school on mathematics education. Throughout this, I’ve continued to have a passion for teaching, helping others learn and helping to scale that impact. As my interests have developed over time from let’s say starting in education to teaching small classes to my time at the University of Minnesota … where I was teaching a lot larger lecture courses … that still topped out at 150, 200 students.
Mike Delgado: Wow.
Eric Weber: It’s big. But when you think about the level of impact, you’re thinking, “Okay, I can affect these hundreds of students.” But there’s still an upper limit on what you’re doing. During the time I was at Minnesota, I had the good fortune to be part of the University of Minnesota’s Business Analytics Program, the first part-time cohort who went through the program. In short, I was just fascinated by analytics, by data science, by the power that you seemingly have at your fingertips, which I think still blows me away when I’m doing work. After I finished going through that program, I had a lot of open space in front of me. The question was whether to stay in the academic setting, where I continue to love teaching. Even being at LinkedIn, I do miss the day-to-day teaching component of things.
But I had the chance to come to LinkedIn specifically to work on their learning product. It’s the newest venture here, the newest product that certainly is a little bit different. But we acquired Lynda.com about two years ago now, so a lot of what we focus on is trying to help professionals develop the right stills. Help them develop the right skills they need to get the jobs they want and develop in the jobs they want and advance their careers. A big deal for me was the chance to do this at scale. I think that’s something we have a chance to do with LinkedIn. In LinkedIn Learning specifically, we have the advantage of having this large network that’s already built, this exchange of ideas between professionals. I think we’re just getting started. That brings me to today, where that’s my focus every day — trying to expand the impact as part of our sales team of Arlington Learning Products.
Mike Delgado: What I think is really cool, Eric, is that your passion for teaching and education and data sciences is merging together for you at LinkedIn. Even though you don’t have a hundred students in front of you, you’re now educating possibly thousands or millions of people just through the products that you’re helping to serve up to the right people. Right?
Eric Weber: Right. I think that’s part of the reason I care so much about what I do day-to-day. This is why I tell people, “Find what your passion is.” Some people are passionate in general about data science and that’s awesome. I am passionate about data science, but specifically about trying to help other people learn, and that opens up unique paths. I think you’ve got to find what you care about because data science is a whole bunch of opportunities. You just need to decide exactly what your opportunities should look like.
Mike Delgado: Eric, can you share a little bit about some of the data science work you do at LinkedIn?
Eric Weber: Generally speaking, I work on LinkedIn Learning and I specifically support our sales team. At the end of the day, if you wanted to describe what I do in a relatively simple way, everything I build and all the insights I try to provide are directed toward making our sales team more efficient, helping our sales team understand our customers better. Right now, we’re building a new business trying to help our sales team find what the right profile is for people we want to bring into the fold. I think that’s a really fascinating part, especially when you’re starting a new venture. You get to see things like customer segmentation that other people might find relatively boring. I think it’s really fascinating because segmentation opens up all these insights the sales team may not have otherwise had.
At the end of the day, I’m trying to help our sales team be more efficient. That can include simple data pulls and insights. It can include building more advanced and scalable machine learning models over time. But sometimes when I go into work, I never know what it’s going to be on that particular day. But as long as it serves the purpose of making the team more efficient, that’s what I do.
Mike Delgado: That’s awesome, Eric. Just over the last year, I’ve seen leaps and bounds in the types of products and things going on within LinkedIn. The communities are forming. I’m seeing great discussions. I feel like LinkedIn has changed for the better. I joined LinkedIn probably a year after it started. It more like a resume-type site. Now, just in the last year, all the different learning modules, the adding of the skill sets and the stream. It almost feels like a business Facebook. I’m seeing really great discussions happening in the stream.
Eric Weber: Yeah. I was similar. I’ve probably been on LinkedIn maybe since 2010, 2011. But the way I think about it has changed pretty dramatically over time. I think the way a lot of people think about LinkedIn has changed dramatically even over the last year or two, and that’s intentional. We want to move beyond this concept of a place where you store your information so recruiters can find you. That’s certainly an important part of our business, but there’s also this community-building aspect to it. I spend a lot of time reading through my feed, getting people’s perspectives on things.
Given the outside world and things that are going on and how Facebook and other social media platforms operate, I think people like to be in a place where people focus on being professional. To me that’s a cool thing today. People have this … It’s not a written rule, but there’s sort of an unwritten rule of “You’re gonna share stuff here. Keep it focused on professional things.” I think that’s an awesome unwritten rule to have.
Mike Delgado: It was because I was reading through the streams that I came across you and your article on the importance of creativity in data science. I want to talk with you a little bit about that. Why is creativity so important for a data scientist to have?
Eric Weber: There are a lot of ways I could respond to this. The most direct reason for me is that when companies think about hiring data scientists, they often have ideas about what they want them to do. They know they want them to be able to organize and pull data. In some cases be a glorified analyst; in other cases build more advanced machine learning models to do particular tasks. But the important part there is the company is so familiar with their data … Well, in some cases they are … that they already have the set of tasks they want to accomplish. In a lot of ways they don’t know what they don’t know. They’re missing out on what’s possible.
Very often if you’re not in this field and if you’re not working in a specific business or type of context day to day, it’s difficult to know what you’re missing out on. Often that creativity part comes from the data scientists themselves. They need to be able to see maybe what’s missing, to pick up on how things could be done differently, to envision what something might look like two or three years down the road, and start that process or data pipeline development early in order to accomplish those sort of goals.
There are a lot of aspects to creativity. One is being able to think about things operating in a different way than they currently do. Of course from a business side, if you’re in a C-level or something else, you don’t dig into the data enough day to day. You don’t dig into the predictions enough day to day to understand how it could be different. But the data scientists can. That’s where they add a lot of value — not just in building models and pulling data, but in trying to actually produce innovation for the business and innovative ideas.
Mike Delgado: To your point, I love the fact you’re talking about just how creativity from the very, very start, from just being curious … How can we change things from what the norm was, from crafting the right questions … to, like you said, what type of data should I be pulling? All that is definitely an art.
Eric Weber: Right. I think that’s the hard part. If you look at people’s questions or what … we have trending topics. But the things that get asked all the time are how to get into data science and what the required hard and fast skills are. Certainly, there is a base set of skills. You need to be able to manipulate data in SQL, Python, R, whatever it might be. You probably need to be able to build some predictive models. You probably need to be able to communicate with other people, but the art part of that is how you combine those different things. Not everybody’s going to do it in the same way. Every particular task or project that you do is probably going to require you to leverage different parts of your skill set, and so it is an art. You can be ready to produce art, but I can’t tell you how to paint something.
People are often frustrated. They say, “I want the solution.” People who are data-focused want a hard-and-fast or a rule-based system to becoming a data scientist and knowing how to do it, when in fact that often depends on a lot of factors that are project-specific or company-specific.
Mike Delgado: Eric, can you share one or two examples of how creativity has helped you in your work either at LinkedIn or previously in products you’ve done?
Eric Weber: Absolutely. I think in the case where I work with our sales team, they care a lot about being able to identify new customers, being able to identify a business they should pursue. I won’t go into a ton of detail, but often there is, on the business side … How do I put it? … They have real ground-level expertise. They’ve developed ideas about what works and what doesn’t over time and what signals are about who they should pursue. Often the data can tell a little bit different story about what those signals are. You end up having these two things. You have maybe the business-side sense of what’s important, and you have the data-side sense of what’s important. The truth of the matter is probably somewhere in the middle.
I don’t think anyone’s necessarily wrong. They’re just seeing different signals. It’s trying to be creative about figuring out, “Okay, you think this. I think this, so I don’t think either of us is necessarily wrong. It’s just a question that we’re probably seeing something different in the data.”
It’s that creativity part, and creativity of course necessitates things like being very driven. You need to want to see these things through to completion. It requires patience, trying to find that true middle. I think that’s true of anyone who’s working in a business context, being able to find that truth that’s in the middle. It often requires creative approaches, not just to problem solving, but to how you communicate your own results with the business side.
I think the other thing goes back to what I said before in trying to innovate, trying to say, “All right, this is how we’ve done things in the past. This is how we’ve organized the sales team. What if we do this a different way? What if we think about our account strategy in this way as opposed to this one?” Trying to do a test, for example, of how effective a particular sales approach is or something like that. For data scientists, the notion of A/B testing is very straightforward for most. They think about it in the sense of a product where some users are served up this new version of a feature while others are served up an older version of a feature and you compare their behavior. But when it comes to looking at things like marketing or sales, the idea of what an A/B test is is a little bit different. Because ideas like randomization and what you’re measuring, they become a little more complicated, and a lot of creativity is required in that sense.
I’ve learned a lot from the other people on our LinkedIn sales teams and analytics teams about how to think about these issues. Those are just two examples. I think it’s probably true every day that I could pick an example where creativity matters, but those are two big ones for me.
Mike Delgado: Eric, I’ve got to say that the first example you shared is really cool. How much respect you have for the business intuition side based on somebody who knows the business and has been in the business for a long time. They’re sharing their insight with you and then you’re looking at the data and maybe coming up with a different point of view.
Eric Weber: Correct.
Mike Delgado: You’re like, “Okay. How do both of these points meet? It’s not an either/or; it’s a yes and let’s find a balance. Let’s see how we can learn from each other.
Eric Weber: Yeah. To me that’s key. When you are in a data-driven position, I think a quick way to get yourself out the door is to say that you know everything. Because in reality I don’t know as much about sales as someone who’s done sales for 10 years. I might know more about the data but, again, I think about the data or what I’m seeing as an end product. I see the results, the sales process, but I don’t know what it’s like always to be on the ground and to be doing these things. There’s a certain measure of respect for the business side. I think this is true regardless of the industry you go into.
I would find it concerning if a data scientist walked in and said, “Here’s the story. This is what’s going on,” without taking the time to assess what the business thinks. To me you are delivering value for what you can add to the business, so it’s silly to ignore what the business side actually thinks.
Mike Delgado: It sounds like humility is a big part of this. Right?
Eric Weber: Yeah. I couldn’t name names about who I’ve read make posts about the importance of humility, but with data I think sometimes you can feel powerful like, “I have all the insight. I can see everything.” When in reality with data, you don’t see everything. You have a pretty limited slice of the actual business. It’s whatever you’re choosing to measure at that particular time. Understanding that you probably don’t have a full sense of what’s going on is important. It doesn’t mean that you’re not good at what you do. It just means you’re being real.
That’s something when I first got into data science, when I was building models, I was thinking, “All right. I have developed the solution. This will solve all the problems.” Probably not. It’s probably not going to solve all the problems, but it might move the business forward a little bit. I think having that sense of I can do something and you’re part of a team effort is more important. When I interview people and I ask them questions about how they would interact with the business side, I look for that. I care about if you have a sense of humility, that you don’t know everything. That’s actually sometimes rare to find.
Mike Delgado: Eric, would you say that part of that humility for you came from academia as you were learning and realizing all the things you didn’t know? Would you say that’s fair?
Eric Weber: Yeah. I think that’s a good point. As an example, when I was at Minnesota I was teaching statistics and about design of experiments and clinical trials in a biostatistics department. My office was next to world-class people who were running these trials in foreign countries during major breaks. When we had the Ebola crisis in Africa, the people who were on the ground doing the clinical trial tests for vaccines were from that department.
That’s really humbling no matter what I know that there are going to be people who have 10 times the knowledge I do. I think it’s a pretty humbling experience. Academia is certainly, what, does it put you in your place? I think probably. While the sales side that I work with, they don’t have such a thing as a tenured professor, but you figure out who the tenured professors are. Okay, this person has essentially ruled the sales world 15 years and they know how it works. There’s that measure of respect saying, “I’m not going to try to tell you that you don’t know what you’re talking about. I’m just going to figure out how to help you wherever I can.”
Mike Delgado: Eric, one last question in regard to art and creativity. I read a stat in Forbes that was published two years ago that said 76 percent of data scientists view data preparation as the least enjoyable part of their job. I’m curious about maybe those tedious aspects of data science. How do you stay creative and curious? How would you encourage someone to be creative during those tedious moments?
Eric Weber: Well, one, I’m surprised it’s only 76, but here’s how I think about it. Everybody wants to build the fancy model. Everybody wants to do something hugely impactful, but you’re only as good as the data you’re using. If your pipeline is not good, if the data transformation process is not good, if your data cleaning process is not good, there’s a lot of issues that come up. Because the model that you generate, the insight that you generate, maybe doesn’t have a strong foundation that it needs. It’s like eating your vegetables. Actually that’s not a good analogy because I love vegetables, but most people are not huge fans. It’s this conditional part that’s going to make everything else you do impactful and reliable. I think it might get overlooked in some cases, but it feels very bad.
Imagine you’re in a sales scenario and you’re telling the sales team something that is surprising to them, or maybe that they don’t agree with, or that goes against conventional wisdom. The last thing you want to do when they say, “Well, do you trust the data that you’re using?” is to be uncertain at that point. You’ve done a good job establishing that pipeline, doing data transformation, cleaning. You’re not too worried. But if you haven’t, that’s the time where things start to feel very shaky. Let’s say that they go into your data, and they say, “I found this issue with what you were doing.” That can poke holes in how much they trust the insights and the things you’re developing.
To me, you’re not going to be able to do the fun stuff without doing this. Sure, 76 percent, 80 percent, whatever it is, may be the least fun thing, but it’s probably one of the most important things.
Mike Delgado: We only have a couple of minutes left, so I just want to ask a couple of quick questions. The first one is this. For those who are watching … those who are listening to the podcast, aspiring data scientists … they want to get into the industry, what would be your advice for them on how to get started?
Eric Weber: I think there’s a couple of things, two points of advice. One is that you need to figure out what your passion is. What do you like doing? If you’re in data science for the paycheck or the title or something else like that, you’re going to burn out pretty quickly. The reason is that the work is enjoyable if it’s the data science part of it you like. If you’re chasing these other external things, I think it can be very difficult for someone to really love that life. You’re going to get tired. Spend time assessing what you care about doing. Do you care about sales? Do you care about marketing? Do you care about security? You don’t necessarily care about the field you’re in, but do you care about building machine learning models? Would you rather be pulling data and generating insights using dashboards? Depending on your interest area, the position can look totally different. Finding what you’re interested in matters a lot.
The second part is be realistic. I think what’s really cool and frustrating about getting into this field is that there are incredibly talented people in it. Having an advanced degree and having tons of training in a program and language and being a good communicator are maybe necessary but not sufficient for landing the job you want. Mostly likely, the job search is going to be mentally tough. I think you’re going to get a lot of rejections. I think I still have a hundred rejection emails sitting where I just stored them in one folder in my email. It’s going to happen.
Be realistic about the timeline and also the position you’re going to get into. Don’t always assume you’re going to walk into a high-level data scientist position in a company because the people who are in those high-level positions may have been there for five years and they may be high-level research scientists who are just extraordinarily good at what they do. You have to be willing to accept what you can get and build your way up from there. That would definitely require being realistic; it requires some humility. I think that’s an important part of the job search process.
Mike Delgado: I love your advice. Eric, one last question for senior leaders who are watching this broadcast. They’re looking to build a data science team. What would be your advice for them on how you hire for data scientists?
Eric Weber: Actually a couple of days ago I wrote an article on LinkedIn where I just put a top 10 list of things I think companies need to assess when they’re hiring data scientists. But I think the first question there: Do you need one? I get data science is a really hot field right now. Companies assume if they don’t have somebody in that space they’re missing out. But doing a needs assessment of whether you actually need to hire someone who’s relatively expensive and may be hard to land in the end is an important piece before you do anything else.
Then it depends on the maturity of the organization. If you have data scientists who are advanced and you have them in-house and they’re okay doing the hiring process, great. But if not, getting the right person in the door for data science is important enough that I think it’s crucial to be willing to go outside for help, to ask other people to bring in a data scientist from another company who maybe does consulting, is able to help you with that hiring process.
I think the last part, don’t undersell the importance of the soft skills on the side. I think when people hear data science, they think technical. But at the end of the day you can get the most technical person and if they can’t sell what they’re producing or they can’t explain it in a digestible way to other people, it’s going to be difficult for them to succeed and be helpful in that company. Those are some of the key things I think about. I didn’t say interview about this language or ask this question. Those are secondary to deciding if you need to do this in the first place.
Mike Delgado: I love that. I want to recommend that everyone follow Eric Weber on LinkedIn so you can catch his articles. You can interact with him there. You can find his LinkedIn profile simply by going to ex.pn/ericweber. That’s a short URL that just redirects over to his LinkedIn profile. Follow him. If you have further questions, feel free to interact with him there. You’ll also find all of his blog posts. I think you actually have a Google Doc linking to all of them.
Eric Weber: I do. Very, very advanced and sophisticated.
Mike Delgado: Which is very helpful.
Eric Weber: Cool.
Mike Delgado: Make sure you follow Eric there. Also, as a reminder, we have this Data Talk every single week where we talk about different data science topics. You can learn more about upcoming chats as well as the podcast by gong to ex.pn/datatalk. Take care and we’ll see you all next week. Thanks, Eric.
Eric Weber: Thanks.
Eric Weber is a Senior Data Scientist at LinkedIn focused on supporting LinkedIn’s sales operations for LinkedIn Learning. He also serves as an advisor to the Master of Science in Business Analytics program at the University of Minnesota.
Eric received his Masters of Science degree in Business Analytics from the University of Minnesota, Carlson School of Management and his Ph.D. in Mathematics Education from Arizona State University. Make sure to follow Eric Weber on LinkedIn to keep up with his articles.
Check out our past and upcoming live video data science chats.