I’m not the first person to say that data science is a buzzword. The “data scientist” label gets thrown around a lot nowadays–everyone from a business analyst to a PhD-level statistician likes to call themselves data scientists. Which–sure, I guess that’s not ideal, but saying “NBA Champion Scot Pollard” is just as correct as saying “NBA Champion Michael Jordan,” so really nothing in the world is sacred and we just have to deal with it.
Anyway, this post isn’t about what I believe a data scientist is, or the R versus Python debate, or some buzzword-laden write-up about the five clustering algorithms a data scientist needs to know.
It’s just a post about how I stumbled into data science and ended up with a pretty cool job where I get to answer interesting questions with data.
When I graduated from college, I knew nothing about data science–sort of. I do remember some kid in my sophomore year Mathematical Foundations class tell me he wanted to become a data scientist and work for a major league baseball team–and that a data scientist was someone who knew more statistics than a software engineer and more programming than a statistician. Which… I guess is true–except I also know less programming than a software engineer and less statistics than a statistician, so how impressive it is to be a data scientist really depends on how you phrase it.
But anyway, I graduated with bachelor degrees in economics and mathematics and I was pretty sure I wanted to get a PhD in Economics. So to improve my resume and make sure I wanted to spend 5-6 years prepping for a career as an economist, I looked into full-time economics research assistant positions. These positions are usually at top-tier universities (because they have that $$$) or at Federal Reserve banks. As a research assistant (RA), you support an economist’s empirical research–so this usually involves data cleaning, exploratory data analysis, regression/IV/fixed-effect/etc. analysis and robustness checking (or in data science speak, something something machine learning something something).
So I got a job with the University of California, Berkeley to research energy cost pass-through rates to understand how carbon taxes would impact manufacturing production and output prices. However, the bulk of the analysis utilized confidential micro data from the United States Census Bureau and the United States Bureau of Labor Statistics. This meant that I had to get clearance to access the data and go to a government data center to work on the research project. Meaning, when I was hired nobody told me I had to be interviewed by a U.S. investigator to make sure I wasn’t actually Jonathan Pollard and that I would spend my days in an empty basement in a federal office building near the National Mall with no internet, no cell service, and no human interaction.
Jonathan Pollard. Wait, no, just Scot Pollard again.
But all things considered, this ended up being a great stepping stone for a career in data science. What I realized was that the parts I enjoyed as an RA–struggling through data and coding challenges, going from messy data to clean insights, etc–I could continue to do as a data scientist at a startup.
Once I decided to make the switch from academic research to industry data science and analytics, I faced two major obstacles: one, figuring out what skills I needed to make this career transition; and two, figuring out how to make my experience and background sound like a good fit on a data science and analytics team.
Figuring out what skills I needed to make this career transition
The skills needed to get into data science and analytics can mostly be broken into three categories: technical skills, communication skills, and business/product sense. In terms of technical skills, how much statistics and machine learning you need depends on the position and the company. However, what I’ve learned is that it is a bad idea for aspiring data science and analytics applicants to immediately jump into trying to learn fancy machine learning or deep learning models.
Most first-time data science and analytics applicants won’t be vying for the job openings that are looking for deep learning expertise for a few reasons: one, most companies don’t actually have the data needed for this type of work, and two, only companies with an established and well-run data team will be able to take on a project like this (and have it make any sort of business sense)–and these companies will be looking for highly-technical and well-experienced candidates. Now, if you’re coming from a highly-technical PhD program maybe you can land one of these jobs–but if you have a bachelors or masters degree and 0-2 years of experience, don’t waste your time pretending you are an expert at deep learning algorithms because, well… you’re not.
Having a strong foundation in the basics of data science and analytics will take you farther when applying–companies are looking for candidates who have strong problem-solving and critical thinking skills–not candidates who can list five clustering algorithms they read about in a book.
Programming skills are obviously very important–but what surprised me when I starting going through the interview process was how important SQL was to companies. Since I came from an econometrics research background, I wasn’t too concerned about my R/Python and statistics skills. But, there was often a SQL part of the interview process. I felt like “Okay, I’m well-rounded enough in my other technical skills, I can just learn a bit of SQL here and there and that will be good enough.” This strategy failed–and I ultimately decided to really dedicate my time to learning SQL after a company told me that they liked me a lot but were concerned about my SQL background. I see this a lot in the candidates I interview. People have really strong Python/R skills but little to no SQL background–and while SQL is easy to learn, it’s still a red flag and you need to try to be as well-rounded as possible when applying.
Figuring out how to make my experience and background sound like a good fit
The more interviews I got, the more I realized there was some trepidation on the side of the hiring manager to hire someone who didn’t come from a business background. This is really prevalent in the data science and analytics field, since it’s still new and evolving. You have people from other industries who take a boot camp or two and are trying to get into this field, you have recent masters and PhD grads applying–and in my case, I was coming from an academic background where my only experience was in economics research.
I think what a lot of people forget (myself included) is that a company is hiring you to be able to use data to help the company’s leaders make decisions and reach their goals (whether that is to grow or make money or further develop a project, etc). It’s not enough to be able to have all these awesome technical skills if you can’t explain or present your work to a product manager or an executive. So I spent a ton of time working on explaining my experience in a way that made sense to a PM or executive. For example, while I was a visiting researcher at the Bureau of Labor Statistics, I researched the impact of carbon taxes on energy pass through rates. I had to make sure and explain my process and the impact of it in an interesting and engaging way–to both a technical and nontechnical audience. I also had to prepare to answer analytical questions in a creative and articulate way.
At the end of the day, data science and analytics is about using data to bring valuable and actionable insights to your organization’s decision makers. It’s about communicating results clearly. It doesn’t matter what your background is. If you’re a great communicator with little technical expertise, you can learn coding and statistics. If you’re a math and programming savant with poor communication skills, you can work to become a better communicator.
I came from an economics background and some of my coworkers came from engineering PhD backgrounds. I know people in the data science and analytics field who came from management consulting or law or marketing. Where you came from doesn’t really matter, age is just a number, inspiring stuff inspiring stuff–just work hard and you can get a job. It’s not rocket science. It’s just data science.