5 minute read

In a week or two I am graduating from Georgia Tech’s Online Master of Science in Analytics (OMSA) program. This post could be considered my ‘review’ of the program, although it is more of a recap and summarization and less of a critique…enjoy!

Why I decided on OMSA

I decided I wanted to get a second masters degree in mid 2018 when I was working for IBM around some incredibly smart people who had so much technical knowledge. I was in a project manager role and had a strong sense of imposter syndrome trying to wrap my head around what it is I was supposed to be managing (I still don’t know what an analytics cube is).

When considering the sources of power outlined below, I tend to think that the ones in the ‘personal power’ category as more authentic and legitimate, and I have always admired those who have strong expertise in technical subjects. I decided I wanted to get more technical depth, and have always enjoyed analyzing data, so figured an analytics degree would be up my alley.

5 types of power
The 5 types of power

I am already a graduate of a 2-year full-time MBA so more full-time school was a no-go. Conveniently, Georgia Tech had a fully-online 2ish year analytics program that was very affordable, so I got together my resume, asked for recommendations, and applied.

I ended up getting accepted to OMSA in fall of 2018 to start in the Fall 2019 term. I was somewhat disappointed since I wanted to start right away, but luckily there was the option to take some courses for credit prior to starting in the program (more on that later).

My OMSA experience: The good, the meh, and the bad

The good

  • Exposure to some of the best professors in the country: Georgia Tech is ranked as the #8 best engineering school in the country; the OMSA program is a combination of industrial engineering courses (#1 in US), computer science and engineering (#5 in US), and business information systems (#5 in US)…can’t beat that
  • Exposure to fascinating content and concepts: This may be applicable to all analytics degrees, but it’s worth emphasizing how much I’ve learned. Before starting, machine learning was more like the movie Smart House; now, I know it’s all just optimization in disguise (well, mostly)
  • The cost: The degree cost me $10,430. Compare that to the other top programs here: Carnegie Mellon ($67k), Cal Berkeley ($73k), Harvard ($35k). It’s incredible that the program is so affordable.
  • The Micromasters option: I had two semesters of waiting between when I got accepted to when I started the program. Conveniently, GaTech offers three OMSA courses via a Micromasters certificate that counts for credit towards your degree. I took those courses before starting the actual program, and scored well enough to get transfer-credit for all three. Taking the courses through the MM is cheaper than taking the same course through GaTech, but there is no guarantee you will get credit (GaTech is weirdly vague about what grade you need in the MM courses to get transfer credit).
  • The Slack community: One of the cons of a fully-online program is no hallway conversations, classroom discussions, etc. OMSA’s equivalent is a very active Slack workspace, with channels for all the classes. While I did not participate very often in the main discussions, I found the Slack very valuable for reading others’ thoughts and getting help in my classes.

The meh

  • I self-taught many things: Without the financial skin-in-the-game, I doubt I would have had the discipline to actually learn these concepts, but in the online setting I was often googling the concepts and learning on my own. A few classes I didn’t even watch the lectures at all, and learned while doing the homework and projects. Comparing to my MBA, where in-class discussion of concepts was a key value-add, I wonder if those discussions even occur in the in-person version of an analytics degree, since concepts are more technical/math-heavy.
  • I paid for a piece of paper: I will soon have three expensive pieces of paper. I enjoy learning, but I realize there are better ways to learn than lecture-homework-lecture-project-exam-grade-repeat. This is (99.99% chance) my last formal degree – the rest will be learning-by-doing with hobbies/personal projects.
  • It is a lot of work: Expected, but worth noting. I spent many weeknights staying up late watching lectures or doing homework, and basically dedicated one of my weekend days to the program. I am lucky to have a great support system which made things easier, but it definitely is not for the faint of heart; expect it to be a grind.

The bad

  • Some courses were flat-out bad: This happens in academia, but is still worth noting. Typos in homeworks, boring/bland lectures, and confusing test questions were not common but existed in pockets.
  • I did not take advantage of office hours: I didn’t once participate in office hours (not a complaint on the program; I never prioritized it), so I felt somewhat disconnected from the instructors. I mostly used Slack channels to ask questions on homework to my peers.


Favorite classes

  • Computing for Data Analytics (CSE6040): The first course I took in spring 2019. Best exam format (24-hour Jupyter notebooks with hidden test-case cells; if your notebook runs end-to-end you get a 100%), very cool concepts
  • Computational Data Analytics (ISYE6740): Aka machine learning 1, very tough assignments but I learned a lot. Did a great job balancing math, coding, and concepts.

Most memorable concept

Spectral clustering: Clustering is often done by distances (how close is a point to the other points). Spectral clustering uses graph theory and an adjacency matrix to properly cluster atypical shapes (see the circles in the below image; source)

Spectral clustering
Clustering techniques

My course schedule

  • Spring 2019 (2 Micromasters classes): Computing for Data Analytics; Data Analytics in Business
  • Summer 2019 (1 Micromasters): Introduction to Analytics Modeling
  • Fall 2019 (2): Bayesian Statistics; Simulation
  • Spring 2020 (1): Data & Visual Analytics
  • Summer 2020 (2): Computational Data Analytics; Data Analysis for Continuous Improvement
  • Fall 2020 (2): Deterministic Optimization; Regression
  • Spring 2021 (1): Practicum (employer-sponsored capstone project)

Biggest ‘OH S**T’ moment

I got my dates mixed up and forgot to submit my final assignment for Data Analysis for Continuous Improvement, which was worth something like 40% of my grade. This was going to delay my graduation a semester and cost money…I was not happy. After throwing a temper tantrum, I messaged the instructor who was very understanding and let me submit late for no penalty. Thanks Lee!

Biggest challenge

I was taking two pretty tough classes in Fall 2019 (Bayesian Statistics and Simulation) while traveling 4-days a week for work. On top of that, my wife and I got a dog, bought a house and moved, all within a 3-month span.

I not-so-fondly remember sitting at the kitchen table doing a Simulation exam on a Sunday night after moving the day before. Boxes were everywhere and my wife was unpacking, but I had to do this exam before 11:55pm that night, and was getting on a plane the next morning. That was tough; I am lucky to have a great support system.