My advice to Data Science students
So you’ve decided to take the leap of faith — you’re going to dedicate a year or two to study the best subject in the world (maybe debatable).
It’s already been two years since I decided to take the leap myself with a data science master’s at the University of Leeds; it made a big dent in my savings, and I was unsure whether or not it was the right decision at the time.
But even now, I still think one of my most valued experiences comes from my data science student days.
I know there’s probably plenty of other data scientists who’ve done the same master’s as me but thought it was a waste of time. The main reason I’d disagree with them is because I made the effort to maximise my time as a student, and it paid off for me in the long run.
And when I say maximise, I don’t just mean doing well in exams. I mean finding other things to do that would expose you more to what it’s really like working as a data scientist.
I’ve seen students who did well in exams but didn’t really see the point in what they were doing. I’ve also seen students that might not have done as well in exams but had a much more diverse portfolio and understanding of what data scientists do.
If I was an employer, unless I’m looking for a research scientist, I’m more likely to be interested in someone that has a flexible understanding of what data science is about, and a diverse portfolio would tell me that more than just stellar exam results.
If the university offers any team-based data science projects that are outside of your usual syllabus, seriously consider applying for them.
They force you to work with people with very different backgrounds, and it teaches you to effectively communicate with them despite this.
It might not be immediately obvious where to find these, so in my opinion, joining something like a Data Science Society would help if the university has one. If not, there’s always challenges that are available online for people to form a team to take part in.
When I was in Leeds, I joined the Data Science Society and found that they were looking for a team to take on the Hiscox University Challenge. At first, I was hesitant and it required a lot of time to submit an application; however, I knew the experience would be worth it so I applied anyway.
I successfully joined the team but the competition was tough. We didn’t win in the end, but the experience alone gave me a chance to be featured on Tableau’s Generation Data series, which would probably never had happened otherwise.
Afterwards, a team member suggested to stay together for another team-based challenge found online. We never got to do it due to exams, but it proves how important it is to work with others since it can really open up opportunities for you in the future.
Ever since then, I’ve had it on my CV and it’s given at least something for me to talk about during interviews. Just that alone is well worth having (especially as a graduate), and interviewers can definitely appreciate that you’ve worked on a real-world data science problem as a team.
Searching for jobs early in your studies will give you an idea of the current market trends are, and will allow you to prepare accordingly.
Make a list of what skills employers are looking for and invest some free time into acquiring those skills.
With that being said, it’s very common for companies to include things in their requirements that they don’t actually need, but are just “nice-to-haves”; don’t feel like you have to learn everything listed, but do pick out the important ones and pay a bit more attention to those.
For the “nice-to-haves”, I’d suggest using the Pareto principle where learning the most important 20% of a topic is enough for you to understand 80% of its results.
Job searching can be a time-consuming activity, and since you’re already dedicating to a year or two to studying data science, you might as well try to pick up some extra skills that will make you stand out when you start seriously applying.
This follows on from the previous section, but I’ll repeat it again — make the most of your free time time as a student by learning new skills.
Of course, exams and coursework will always come first, but if you notice you’re not learning some of the skill requirements you’re seeing in job postings, it’s best to organise some time to start learning those as well.
For example, if you notice a lot of job postings are asking for Tableau but you haven’t used it at all yet, I would create a Tableau Public profile and start publishing your work. You could easily find an old piece of coursework that you did and create something that tells a story so that employers can see that you actually have some experience with it.
I think it’s quite easy for students to fall into the trap of focusing too much on the theory and often end up overcomplicating a problem. Sometimes, it might really be the case where you need to use Jupyter Notebook with the most advanced machine learning packages, but a lot of other times a simple Tableau dashboard and SQL query is probably enough.
This is especially true for a lot of companies that already use these tools, so having this foundation will make you more attractive to them when you start applying.
Depending on what you’re learning on the course already, I think some of the following would be good to know more about:
- Data visualisation tools like Tableau or Power BI (a lot of companies already pay for reporting tools).
- Data querying languages like SQL (not as “sexy” as Python but is very commonly used).
- General languages like Python or statistical languages like R (choose the one you don’t know yet — you wouldn’t be a real data scientist without one of these, right?).
- How to write optimised, production-level code with unit testing (this is often overlooked by data science students).
- Cloud architectures like AWS, Google Cloud, or Microsoft Azure (a lot of start-up companies have some form of cloud architecture).
- Big data systems and how SQL/NoSQL databases are structured within them, as well as ETL processes (helps with understanding where and how your getting your data).
- Statistics and how to apply it in a business (things like A/B testing, measuring uncertainty, etc.)
There could be many more, but I think the point is clear: data science is a very broad field, and it pays to have a good understanding of a wide range of topics. Plus, the learning never really stops even after you begin your career (and in most cases, I’d say it probably increases), so you might as well get into the habit early.
After all the exams are over, it’s time to think about what your thesis topic will be. At this point, it’s understandably tempting to pick a topic that sounds interesting but is also something that your comfortable with.
However, I think this is where early research can save you a lot of stress. If you’ve had some time to job search and learn various tools, you’ll find it a lot easier to pick a topic that you know you’d enjoy doing, but also pick with confidence knowing that certain employers would find interesting.
At the end of the day, you don’t want to spend 3 or 4 months doing a research thesis that doesn’t help you with the direction you want to go career-wise, especially if you don’t enjoy the topic.
As an example, I’ve always had a fascination with natural language processing (NLP) and sentiment analysis, especially with online text data. I also had a hunch that a lot of companies probably had more text data than they did any other type of data.
Since I already had some NLP experience from previous modules, I decided early that I wanted to pick or create an NLP topic for my thesis where I’d mine my own text data online.
This saved me from a lot of stress immediately after my exams, and it allowed me to better prepare for what type of content I wanted to cover in my thesis rather than stressing out over which kind of topic to choose.
Although studying for a master’s can be stressful at times, they can also be an extremely rewarding investment. Truly take the time to plan ahead and do your own research about what’s out there — doing so will allow you to fully maximise the freedom and flexibility that a master’s can offer, and a lot of employers will definitely take notice.
Afterall, master’s degrees aren’t cheap these days, so why not make the most out of it?