Was it Worth Studying a Data Science Masters?

As I started to apply for Data Science roles it quickly became apparent that I was lacking two key skills: applying Machine Learning and coding

By Sterling Osborne, University of Manchester

Since completing my Masters in Data Science, I have had a number of people contact me asking for my experience with the course and whether it is worth recommending. Therefore, I thought it best to summarise my decision for starting the course, what I have achieved during my studies, and the outcome in the years following.


Why did I choose to study Data Science?

It was the spring of 2016 and I was coming towards the end of a 6 month internship at one of the largest consulting firms in the City of London. I had taken this role to gain experience and figure out whether becoming an Actuary was the correct route for my career. I quickly found passion in the data analytics of the role as I was being pulled into meetings to discuss numbers I had crunched or was able hack together a tool to automate previously manual tasks. However, I also found that the traditional track I would be heading on if I moved onto the graduate scheme no longer interested me due to years of standardised exams and little to no creativity. Furthermore, most of my work at this point was within Excel and I was gaining little to no coding experience.

It was at this point I also started to explore a magical term that kept coming up in my job searches: Data Science. I had come from a background in mathematics and, due to the nature of the job market in the UK, had been nudged towards the well-founded traditional roles such as accountancy and actuarial consulting. And yet, here was a new role that defied all the expectations I had set for myself of my future career. Where becoming an Actuary I would be solving problems by learning regulatory standards, Data Scientists are encouraged to creatively find solutions that fit within the commercial environment. Furthermore, the role opportunities were no longer fixed into a few select firms but almost all companies were looking for some variation of a Data Scientist and the idea that I could move into a completely new industry, from fashion to finance, greatly appealed to my interests.

However, as I started to apply for Data Science roles it quickly became apparent that I was lacking two key skills: applying Machine Learning and coding.


What were my options?

The first solution was to simply teach these myself, I had the statistical experience to learn machine learning and had done enough coding in MatLab to feel confident that I could learn Python or R. However, if I was to do these I was unsure if this was enough and how I would clearly demonstrate my newly acquired skills on a CV to employers.

I considered online boot camps but they were often fixed in their content and I was unsure of how repeating someone else’s work would be received by employers. Furthermore, there was no guarantee that these were credible to employers and were expensive to self fund. Today, only three years later, the list of courses supported by universities makes this much more of a viable option and is something that is definitely worth considering but at the time these were lacking. Unfortunately, boot camps were a costly risk that I was uncertain would pay off.

Therefore, I decided to look for the options available at universities. At the time there were two types of courses that fit within my goals; business analysts courses and computer science machine learning. The former focused on applying analytics within commercial environments but, as this was run through business schools, was far more expensive at over £25,000 for one year of studying. The latter provided the teaching through academic research and focused more on teaching the underlying theory than the application. Furthermore, as this was an academic course run through the Computer Science department, the cost was considerably less for the year at £9,000 (for UK citizen).


What did I choose?

In the end, I decided to play it safe and commit a full year to the academic masters in the hope that I would gain the applied knowledge in machine learning and develop my coding skills through project work.

Therefore, I joined the 2016/17 cohort of the Data Science MSc at City, University of London’s Computer Science department. This was the second year the course was offered at this university and, at the time, was the only university in London that offered a Data Science Masters (though others had some variations of this).


What did I learn in the course?

The first term consisted of the three main topics of Data Science: fundamentals of data science, machine learning and visualisations. Each module consisted of a coursework component that we were given a choice of any publicly available dataset to apply our newly learned methods on. With these, I was quickly able to improving my coding skills and even built the confidence to start sharing these projects publicly.

In the second term we had two core modules, Big Data and Neural Computing, and were given the choice of two optional modules. The list of options was comprehensive and enabled us to pick specialisms from computer vision to data architectures. I chose Data Visualisations (a continuation of the first term’s module) and Software Agents (the basics of AI by applying Reinforcement Learning). Again, these modules included coursework and with the fundamentals from the first term, I was really able to expand my applications and think creatively. Big Data also introduced text data and natural language processing.

Over the two terms, I had been given a broad overview of most of the Data Science topics and had a deep knowledge of Machine Learning and Neural Networks from the core modules. As we moved into the first component of the course, the dissertation, we were given the choice to complete this whilst in an internship role (and be provided an extension on the deadline to account for balancing whilst working). I found a suitable role, defined my research topic and over the following months applied all the skills I had gained so far towards applying AI within business.


Did I gain the skills I needed from the course?

I had two goals to achieve; to demonstrate that I understand machine learning and apply these with coding. The course not only provided a clear ‘box tick’ against these on my CV but enabled me to continue to expand my skills after with more and more interesting projects. Part of any job application is to get past the initial checks and I was now doing this much more consistently. Furthermore, as I moved into interview stages, I had all these project to discuss and truly demonstrate confidence in my understanding far more than I would have achieved on my own.

The Masters opened up all the doors that I had previously been knocking on and even had recruiters contacting me directly following the projects I had posted publicly. In the end, I found that I enjoyed the research aspect of my dissertation and the freedom to pursue the field and have since move onto a PhD in Artificial Intelligence. Ironically, this is the last thing I would have considered back in 2016 but as the field is constantly expanding it is incredible to be at the forefront of this, particularly because many problems require an applied mindset and fit within commercial problems and are not simply theoretical.


What advice would I give to someone considering moving into Data Science?

This is always a hard question for me to answer as each person’s position is different so will try to offer some advice based on my experience. In short:

  1. Establish that Data Science is the right route for you and find a topic or detail to motivate this. For me, it was the ability to apply data analytics in a creative way and become a valuable asset in a business to help others improve their decisions.
  2. Evaluate what the jobs you want are asking for and where your set of skills currently fall short. Although I had the technical background, I had not demonstrated my ability to apply machine learning or code and needed something to achieve this. For others on my course, they had not come from maths or stats backgrounds and so needed these to strengthen their knowledge.
  3. Review the options available to you to gain these missing skills. For example, boot camps are increasingly more credible in the industry but they are often following a single path (i.e. learn on the same data and apply the same methods). This may work for some but I wanted to demonstrate my abilities in unique ways to stand out to employers. If you are considering a masters, thoroughly research the modules and who is organising the course as variations will exists between different departments, particularly business and academic schools.
  4. Expand and challenge yourself through the course, don’t simply pick a course that covers topics you are already comfortable with to tick a box. Find something challenging that will encourage you to develop new skills.
  5. Demonstrate your new skills by publishing your projects online, either through GitHub or Kaggle or your own site. Building a portfolio of project has taken me further in any interview than trying to describe these within the time limit would allow.

To give an example, below you can find a link to my own website and Kaggle page:

Philip Osborne Data:
This project was created as a means to learn Reinforcement Learning (RL) independently in a Python notebook

osbornep | Kaggle

I hope you find this post useful if you are considering how to get into data science and will do my best to answer any questions you have about this.



Bio: Sterling Osborne is a PhD Research Student in Artificial Intelligence at the University of Manchester (UK). You can find him on Medium as well as YouTube.

Original. Reposted with permission.