Lessons Learned in Pursuing a Master's Degree

Huge life achievement alert! After 4 years of studying, coding, writing, and analyzing, I have finally completed my Master of Science in Predictive Analytics (i.e. data science). It's been a rollercoaster of a ride complete with excitement, growth, perseverance, and pride for a job well done. This cycle repeated itself throughout every course, but also over the course of the entire program.

On three separate occasions recently, a friend has asked to pick my brain about my experience and I thought others who are considering a Master's program might find my insights of interest as well.

Master's Program vs MOOC

Whether to pursue a Master's degree or not depends on your goals, and it certainly isn't a requirement to be a data scientist. There is a lot of amazing content on Coursera, DataCamp, Udemy, etc. for a lot less money. But I've found you only understand how to do data science through those platforms. You don't always learn the underlying math and assumptions that modeling techniques rely on, the how behind machine learning algorithms. Without that second piece of the puzzle, you are somewhat limited in your ability to tweak, customize, and understand your model. So, if you want to be able to fully control your model, then a degree program was very helpful. If you are satisfied with out-of-the box models from tools like Alteryx, DataRobot, AWS's Machine Learning Service, etc. then you can probably skip the formal education.

Personally, I started off using the above references and did that for about 4 years before I needed something more substantial. I understood most of the basics, but was looking for something more holistic to formalize my understanding. Something with a clear curriculum that took me beyond what was freely available online. Something that would motivate me to work towards something bigger.

Choosing a Master's Program

After making the decision to pursue a formal education I started researching the various programs available and thinking about what I wanted in a program.

Do I want to meet online or in-person? At first I thought in-person because I get distracted easily and I wasn't sure I would be able to stay motivated enough to see it through, but in-person limited the programs available to me, so I kept an open mind.
What are the application requirements? It may sound ridiculous, but I really didn't want to have to take the GRE in order to apply to a graduate program. Some require it, others don't.
What does the curriculum cover? By comparing different programs I figured out what I wanted to learn and it wasn't data modeling, ETL, or warehousing. I wanted to focus completely on the algorithms and the mathematics behind them. Every program is different and in many cases you can customize the curriculum with electives or specializations.
How long has the program been around? It was important to me that the program be reputable and have a solid understanding of the material they were teaching me. Many data science programs are so new they have only been around for a few years. I wanted an organization with a track record.
Who are the faculty members and are they legit? I wanted to be sure I was being taught by people with experience and expertise. It was exciting to read some of the bios of faculty members I might learn from and see all the exciting ways data science can be used.
What does the program cost? Of course is going to be a factor. I was lucky enough that my employer helped to cover some of the cost.

Time Management

Once I chose a program and applied, it was soon time to start with the coursework. My particular program followed the quarter system with 10-week courses, 4 times a year, and a 2- or 3-week break between quarters. My initial thought was that I would take 1 course at a time, since I was working full time, and take every other quarter off. It seemed like a reasonable pace, until I realized it would take me 6 years to complete the 12-course program! Ultimately, I decided to take 1 course each quarter and take breaks as needed, which I did only once.

Over time I developed a pattern. It started with being really excited for the course to start. A few weeks into the course I'd be inspired by all the information I was absorbing, models I was building, and concepts I was putting into practice that all my free time was spent on coursework. By week 8 of the course I was calculating what my grade would be if I didn't complete any future assignments. And by the end of the course I was contemplating taking the next quarter off. It was exhausting, but after the 2-week break I always felt refreshed with a new energy.

Here are the time management strategies that got me through:

Prepare. Before a course started I spent a night or two preparing. I read the syllabus, looked through the online portal for the course (becoming familiar with its structure and contents), and even thumbed through the textbooks.
Schedule. I spent roughly 20 hours a week on coursework and I didn't want that time to be spent entirely on the weekend. So I made sure to spend at least 2 hours a night on coursework. From 7-9pm. These small chunks of time allowed me to focus on small, clear tasks.
Routine. The first few courses were difficult, as I hadn't yet figured out my routine. Eventually I realized I could not read textbooks during the week, after the workday was done and I was already tired. So I read as much as I could on the weekends, during the day when I was well-rested.

One Last Tip

Organize your work in a clear structure so you can always find old code you want and use GitHub! You're not going to do it later and you'll be incredibly proud of your portfolio when you're done.