Other

Since data science is such a new profession, there is a lot of confusion and uncertainty around job requirements and qualifications. A common question that gets asked is: what kind of qualifications are necessary to get a job in data science and is a Masters degree or PhD really worth it?

I wanted to share my own experiences with getting a Masters degree in statistics and whether I felt that it has been a necessary requirement for my career in data science.

Why I decided to get a Masters degree

I always knew I would pursue some kind of tertiary education. I wasn’t pressured into it by my family, but I considered it to be an unspoken prerequisite to success. I believed that the only way I could have any kind of success in my life was if I got a degree. Which degree I would get was an entirely different question and I never really knew the answer. I changed my mind A LOT throughout high school and in university. I could never just make a decision and stick to it.

The truth is, I didn’t know where my life was going and I had no idea how to get there. I once dreamt of being an egyptologist, discovering lost sarcophagi and dancing through unseen traps on the floor – or was that just the latest Lara Croft movie? I was an impressionable kid, okay, don’t judge.

Things changed for me a few years out of school when I discovered statistics while pursuing a degree in accounting. Things just started clicking into place, which was a nice change from failing accounting. So I changed my university program and made a commitment to getting a Masters degree in statistics. In no way was this easy and it definitely did not ‘come naturally’ to me. I worked harder than I ever had for anything in my life but it was one of the most rewarding experiences and I’m glad I did it.

To be clear, this goal was arbitrary. There was no expectation or requirement placed on me to get a Masters degree. I believed that it was a high enough qualification that would open doors for me in the future, giving me the best possible chance at a successful career and, by extension, a successful life. I had no idea if this assumption was correct and I didn’t really have any anecdotal evidence for why a Masters degree would be beneficial.

I did not even know that ‘data science’ was a thing when I was in university and I didn’t pick my degree based on a job title. You don’t even need a Masters degree in data science in order to get a job in data science. Many professional data scientists have degrees in other areas of STEM and some are even completely self-taught. At the end of the day, you need to choose your own path.

Upsides to my Masters Degree

So what did the masters degree actually get me? I would say, in large part, it gave me the ability to be independent.

When I was in school, and even in my undergrad at university, my learning was linear and one-sided: study chapter 2 for the test, do these mock exams, solve these problems, etc. I didn’t really have to figure anything out for myself and if I happened to study and remember exactly the right things that appeared in the exam, well, then I was successful. That is nothing like real life.

The final requirement for my Masters was a dissertation that had to be original work with an application that had never been done before. The entire process required to produce this dissertation was intense and one of the most difficult things I had to do at the time. This is the part of my tertiary education that taught me some valuable skills that have transferred into my career and adult life.

The process of preparing my Masters dissertation taught me some very useful skills:

• I learnt how to do independent research and figure things out myself
• I developed resilience to achieve the goals I set for myself and to not give up when things got tough
• I learnt how to write about my research and the results
• I learnt how to select and fit the right statistical model for my data
• I learnt how to prepare and give presentations about my work

However, this only really forms a part of what is required in a data science or analytics-related job. Some data science job descriptions are so different to each other, it’s a wonder they have the same job title.

Each company has unique requirements for a data scientist and the job could lean heavily on either programming, database management and data engineering, business intelligence, machine learning engineering, or even just be highly research-based. Some companies just list all of the above in the job description hoping to catch the Loch Ness monster.

If the variety in the skills required for data science jobs overwhelms you then I’d recommend that you identify 1 or 2 companies that you’d like to work for and do some research into what kinds of skills they expect their data scientists to have and double down on that. This will help to narrow your efforts down to just a handful of skills that you can demonstrate your proficiency in. Companies are much more interested in whether a candidate can actually carry out the duties of the job that is unique to them rather than just being able to do a little bit of everything.

For me, I got my first job in business intelligence and analytics before I even graduated with a Masters degree through a referral by my professor. I went from working on my Masters full-time for a year to then working on it part-time for the next few years while I built up some industry experience. However, the fact that I was working on a Masters gave me a competitive edge against the other candidates who were also applying for the job.

Downsides

Choosing to get a Masters degree and even choosing which university to go to can have many downsides. These are a few of them:

• Cost is probably one of the biggest factors to consider. Most postgraduate degrees around the world are notoriously expensive. However, I was very fortunate in this regard as I was able to get multiple scholarships that contributed towards the majority of my tuition. I also did tutoring for undergraduates that helped pay for my living expenses.
• Skill lag. Most university programs tend to lag behing industry and you may not be learning cutting edge algorithms and the latest developments. While I extensively studied a number of statistical models, we did not even touch on some of the more advanced machine learning algorithms and their applications.
• I was restricted to only the professors at my university and the respective areas they specialised in. This is a factor that was completely outside of my control. I could not pick and choose who would teach me and while the professors may be highly regarded researchers, their teaching methods did not always click with me.

Self-Study

I have had to do much more of my own self-study to learn data science-specific skills. However, there are hundreds of resources out there on just about any topic in data science, machine learning, business intelligence and artificial intelligence. This can be both a blessing and curse.

You can learn as much, if not more, than what is taught in university by using the resources available online and it wont cost you nearly as much. However, the downside is that you are spoiled for choice which usually means that you make no decision at all. It is much easier to bail out of learning anything when you have so many options in front of you since you don’t know where to turn, where to start, which resource is best, and who to go to for support. This is where university programs can be really beneficial – they offer structure, curriculum, and support all bundled up nicely.

However, it is possible to create a structure and a curriculum to self-study the topics and areas of data science that interest you the most. In this way you have all the control and flexibility to choose exactly what you want to learn and the direction you want your studying to take. I understand that this path is not for everyone and you need to find out what works for you.

Self-study is very difficult and takes a lot of intrinsic motivation. There are no deadlines, nobody breathing down your neck to finish a chapter or complete a project. You have to be driven and dedicated to your own study and growth.

If you’re interested in this self-study route, then I recommend you give this blog post a read – I developed my own 6 month data science curriculum with some of the best books and courses in machine learning. There are also a couple of people who have developed and shared their own curriculums as well as step-by-step processes for how to go from absolute beginner to mastery in data science. Check them out and draw inspiration from their journeys.

Here are a few to get you started:

• Daniel Bourke
• Tina Huang
• Ken Jee
• Siraj Raval

To sum up, I believe that getting a Masters degree allowed me to kick-start my career in data science and analytics and I don’t think I would be where I am now without it. That being said, the education I received during my Masters was definitely not sufficient for keeping up with the growth in the field and a lot more self-study is needed so that my skills remain relevant.

At what point do you get to declare “I am a data scientist”?

How many PhD’s, decades of experience or deep learning models do you need to have under your belt before you feel like a ‘real’ data scientist?

The truth is, data science is advancing so rapidly that it really is impossible to keep up with everything. All you have to do is take a look at Arxiv to know that there is a constant stream of new research and discoveries coming in, showing no sign of slowing down and if anything, will only speed up.

So even if you get a PhD or spend 10 years mastering a few algorithms, chances are your skills will become obsolete relatively quick. The key is to develop a thirst for knowledge, keep learning and then solve real problems with your knowledge.

It is unrealistic to expect yourself to truly know it all. This is the unachievable. Instead, to protect your own sanity you need to learn to manage the impostor that lurks within yourself and indeed, within all of us.

You’re not alone

The first thing to emphasise is that you are not alone.

Impostor syndrome is felt by almost everyone, even those that you may view as experts. This is because the amount of impostor syndrome you experience can increase as your knowledge increases. As a complete beginner, you don’t know what you don’t know but as an expert you become more and more aware of all the things you don’t know.

The only thing that sets successful data scientists apart from everyone else is their grit and determination to carry on learning and taking action no matter how much imposter syndrome they feel.

Social media and the internet as a whole allows people to only show their best moments, their greatest ideas, or their most polished projects. So take what you see with a pinch of salt and remember that we all go through feelings of imposter syndrome and self doubt. What’s important is the action you take despite these feelings.

Celebrate your wins

Data science is a complex, multi-faceted field with many areas to specialise in which can be further broken down into each of the domains that their techniques are applied to. The enumerations are endless.

Because of this, your learning is not a straight path, it is a life long journey. You will have many ups and downs but it is important to celebrate even the smallest wins along the way as your progress, step-by-step along your path.

Avoid comparison-itus

Comparing yourself to others (especially those that have “made it”, so to speak) is like catching a bug – once you’ve got it it can be difficult to shake.

While comparison is a natural part of the human experience, you should not let the success of others dictate the way you see yourself. You cannot compare your first step or your 10th step to someone elses 100th step.

Your journey is unique because it is YOUR journey. Nobody else can walk the same path as you and your experiences are completely unique to you. Embrace and love where you are, celebrate others for their achievements and then take the next step on your own path.

Writing this post today feels as much like a letter to myself as it is a message to you. Impostor syndrome will never go away, but it can be managed and it should never be allowed to hold us back from achieving our dreams.

For me, in the past I had always been hesitant to call my self “data scientist”. It sounds so grand and official and I always thought I wasn’t ready yet or I hadn’t learned enough yet. Ultimately, there is no one standing around waiting to bestow a grand title on your name. It is up to you.

Heed the wise words of Brandon Rohrer on KDnuggets:

“Our goal isn’t to accumulate answers, but to ask better questions. If you are asking questions and using data to find answers, YOU ARE A DATA SCIENTIST. Period”

Brandon Rohrer