GTOMSA Course Review
Originally published as two blogs on LinkedIn in May and August, 2021. This has been combined, updated and edited with additional reflections since then.
I recently graduated from the Georgia Tech Online Masters in Analytics (GTOMSA) program! If you want to see my thoughts on the program holistically, check out my blog here: Curious about the Georgia Tech Online MS in Analytics? | by Riesling Meyer | Jun, 2022 | Medium
This blog is great for:
- People considering applying to the GTOMSA or curious about what you can learn in analytics courses in general
- People in the GTOMSA program, and need to figure out what class to take next
- Anyone who is wondering what I’ve learned over the almost 3 years I spent in school while working full time!
As I have mentioned before, these are my opinions so take them with a grain of salt. Some of my favorite classes have been my friends least favorite classes and vice versa! You will likely have different opinions if you have different interest or background than I do. Also, these classes have probably changed; the GTOMSA program takes feedback very seriously and are continuously iterating to make the program better. For aggregated difficulty, workload in hours/week, rating, and other reviews see OMSCentral or omsa.ga (the courses tab).
In this blog, I will review the following courses: Computing for Data Analytics (CSE 6040), Intro to Analytics Modeling (ISYE 6501), Data Analytics for Business (MGT 6203), Business Fundamentals for Analytics (MGT 8803), Data & Visual Analytics (CSE 6242), Simulation (ISYE 6644), Data Analytics and Continuous Improvement (MGT 8823), Regression Analysis (ISYE 6414), Time Series Analysis (ISYE 6402), Digital Marketing (MGT 6311), and Applied Analytics Practicum (ISYE 6748)
CSE 6040 Computing for Data Analytics (Fall 2019)
OMSCentral: Difficulty 3.21/5; Workload 9.75 hours/week; Rating 4.44/5
Computing for Data Analytics is a great course. Each week, you learn something new about modeling, data collection, data preprocessing, data pipelines, or data visualization, and you build many of them from scratch (ie no packages) in python. Some data topics we covered were pandas, SQL, sparse matrices, and compression. Some models that we covered were regression, classification and k-means clustering.
The homeworks were very challenging and 50% of the grade. The other 50% was from 3 exams.
I’ll be honest, this class was hard. I knew basically 0 python going into it, but I had “coding proficiency” in java in college, so I understood coding concepts. I found a friend of a friend in the course, and we met up weekly at a Starbucks for about 4 hours to go through where we each got stuck on the homeworks. I’m not sure I would have gotten an A in the class without her! We ended up taking most classes together, and helping each other a lot!
ISYE 6501 Intro to Analytics Modeling (Spring 2020)
OMSCentral: Difficulty 2.88/5; Workload 10.07 hours/week; Rating 4.25/5
ISYE 6501 Intro to Analytics Modeling is similar to CSE 6040 Computing for Data Analytics in that you learn how to implement a bunch of models. The key differences are this class is taught mostly in R (not python), and it teaches you how to implement packages for each model (instead of coding from scratch). In this class we learned classification (clustering and SVM), data preparation, basic time series models, regression, tree-based models, variable selection, basic simulation, and how to deal with missing data.
Also similar to CSE 6040, this class was hard. I knew basically 0 R going into it. This time I had two friends in the class, and we met weekly for about 4 hours to go through where we were stuck in the homeworks. I learned so much from them!
MGT 6203 Data Analytics for Business (Spring 2020)
OMSCentral: Difficulty 1.9/5; Workload 4.51 hours/week; Rating 2.3/5
MGT 6203 Data Analytics for Business, or as I like to call it, “how to apply linear regression in R to many areas of a business”. This class was useful, a great primer for ISYE 6414 Regression Analysis, but it was really, really easy. It had short lectures, multiple choice homeworks, and easy exams. I’ve heard they’ve made it slightly harder since I’ve taken the class, but it’s still an easy course.
We learned linear regression, indicator variables (i.e. categorical variables in regression), interaction terms (i.e. two variables multiplied together), transformations of variables, logistic regression, and how to use regression to analyze experiments. Then we learned about finance, marketing, and operations management, and how to apply regression in each scenario.
MGT 8803 — Business Fundamentals for Analytics (Summer 2020)
OMSCentral: Difficulty 2.95/5; Workload 7.72 hours/week; Rating 2.43/5
This class is known as the “mini-MBA”. It covers Accounting, Finance, Supply Chain, and Marketing, with an optional unit on Business Strategy. If you have an undergrad business degree, or have taken a college level course in all 4 areas, you can get a waiver for the class. Each unit is completely self contained, with their own homework. There is a midterm covering the first two topics (accounting and finance — 40% of the grade), a non-cumulative final covering the other two topics (supply chain and marketing — 30% of the grade), and 3 homeworks (30% of the grade).
Finance and Accounting: These modules were by far the hardest. There was a lot of material, a lot of formulas, and a lot of memorization. But, it was extremely useful. I can now confidently read financial statements (balance sheets, income statements, statement of cashflows); calculate net present value (NPV); understand amortization; calculate internal rate of return (IRR), payback periods and weighted average cost of capital (WACC) of various scenarios; and, in turn find quarterly earnings calls much more interesting. The homeworks and exam were very challenging.
Supply Chain and Marketing: These modules were substantially easier. In supply chain we learned basic strategies like the news-vendor model, the bullwhip effect, and centralized vs decentralized supply chains. The marketing module was honestly my first exposure to marketing — I used to think of marketing as advertising, but this module exposed me to how integrated marketing can and should be at a company. I learned about the four Ps: product, pricing, place, and promotion, and how marketing isn’t just about selling an existing item, but building demand and working with product development to make sure the product fits the market. We did a laundry detergent case simulation where we could change the product and modify the price and promotion spend. I learned a lot about marketing and supply chain, and the homework and exam were easy — the best of both worlds.
Overall, I loved this class! It was a great overview of business, gave me a holistic view of how companies operate, and helped me understand quarterly earnings calls/reports. I learned a TON, which is all you can really ask for!
CSE 6242 — Data & Visual Analytics (Fall 2020) AKA Riesling’s Declassified DVA Survival Guide 😜
OMSCentral: Difficulty 3.23/5; Workload 15.53 hours/week; Rating 3.01/5
Ah DVA — well known to be the hardest required course in the program, and boy did it live up to its reputation. Not only was the content was difficult, but also homework questions had typos, setting up all of the different programs and environments was a huge pain (the only time I have yelled at my boyfriend (now fiance!) in our 6+ years of dating was when he asked me to go for a walk 4 hours into trying to get Docker to work on my machine), and the project requirements are more tedious and restrictive than they need to be.
That being said, now that it’s over (hindsight is 20/20, am I right?) it’s probably the most important class in the program. You learn/use python, python Flask, APIs, tableau, javascript/D3, SQLite, pyspark, google cloud, AWS, docker containers, and probably more. It was a whirlwind.
More importantly than the tools you learn is the analytics mindset and experience. Most people in the GTOMSA program are doing this for a career change, so this is their first experience of doing an analytics project. You learn how to try and pickup new tools, how to work with and clean real datasets (as opposed to the nice, curated data you get in most courses), how to multitask, how to collaborate on an analytics project, and how to use industry tools for free that might be too expensive or too intimidating for someone to try on their own. Of all classes in the GTOMSA, this course teaches you how to be an analyst.
The course grade consisted of four multi-week homeworks (50% of the grade) and a semester long group project. (50% of the grade)
- Homework 1: Using a public API to pull data, visualizing it, SQLite, D3 intro, and Python Flask intro (a good gauge to see if you’re ready for the class)
- Homework 2: Tableau intro, and a whole lot of D3 (D3 is HARD if you’ve never seen it before, but this is the last time you’ll have to use it — so just get through it)
- Homework 3: Pyspark using docker, data bricks, AWS, GCP, and Azure (the hardest part of this homework is all of the setup!). This homework is more to expose you to tools than challenge you from a technical perspective
- Homework 4: Python heavy — page rank algorithm, building random forest classifier, and scikit learn
- Group project: The prompt of the project is basically “use a big dataset (ie not something you can load into excel), create an interactive visualization incorporating a non-trivial algorithm, and write a paper about it”. Check out my group project poster video here!
Riesling’s Tips to Survive DVA:
- FIND A GOOD GROUP. I repeat, FIND A GOOD GROUP!
- Why? You’re going to spend dozens of hours with these people, and relying on them for 50% of your grade.
- How? Before the course started, I got two friends to take the class with me. We knew about the skills needed for the project from friends who had taken it previous semesters, course reviews and slack channels. So, we met up and created a write up about our group, what skills we excelled at, and what skills we lacked to post on slack and attract other talent in the class. We reached out to 1 person who wrote up his own bio, and had a bunch of people reach out to us so we got to pick who we wanted to join.
- My experience: We had so much fun. 5/6 of us took Simulation together the next semester, and 3/6 of took regression the next semester. Find a good group! It will make this class more tolerable, and it’ll pay off in future semesters!
2. When picking your topic, ENSURE THERE IS A GOOD, CLEAN, USABLE, FREE DATA SET TO USE
- Why? You only have like 8 weeks for the project, and you cannot change your topic after your proposal is accepted. You don’t want to spend the first half cleaning the data. Or, even worse finding out your data set isn’t usable for the problem you want to solve or finding out you’ve hit an API limit and it is no longer free and needing to find a new data set
- How? Spend the time up front before submitting your proposal playing with the data. More time upfront will save you a major headache later
- My experience: We used census data (easy), but we also wanted to add in grocery stores with a google API. We realized the google API stopped being free, and, we had no way to join census data (listed by census tract) to google API data anyway! In the end, we used the foursquare API and joined on GPS coordinates using a python package to convert census tracts and addresses to GPS coordinates. We survived, but it would have saved us a headache if we had considered these impediments before committing to adding grocery stores to our project proposal.
3. Take this class by itself.
- That’s pretty self-explanatory. This class is going to take up a lot of time. Don’t overload yourself.
4. Game plan your total points.
- Your project is worth 50% of your grade. Do that well. Know the topics of the 4 homeworks ahead of time and be okay with bombing one. I only spent 2 hours on Homework 4 and got a 15% on it, but still got an A in the class! It wasn’t worth the stress and effort to get a higher grade on the homework, and it wasn’t worth the risk of not having time to finish the project.
5. Don’t be afraid to take time off from work
- Doing a masters program is hard. You’re going to spend 10–20 hours per week on this class. Take a vacation day every once in a while to catch up or veg out.
ISYE 6644 Simulation (Spring 2021)
omsa.ga: Workload: 10.1 hours/week
Simulation is taught by Dr. Goldsman — he’s sort of a GTOMSA legend. He hands out extra credit like candy and really dislikes Justin Bieber. He’s the most enthusiastic professor I’ve had so far in the GTOMSA program, which makes learning a lot more enjoyable, and he’s the only professor I’ve seen who edits special effects into the lectures!
In addition to the great professor, the class material is really interesting and (mostly) useful. The class consisted of weekly multiple choice homeworks, 3 exams, and two simulation projects.
For my first project, I simulated a dice rolling game in Python. Basically, two people take turns rolling a die and take coins out of a pot or put coins in depending on the number they roll until they run out of coins. Each set of player turns was a cycle. I simulated the base game and two extensions with different starting coin amounts for each player to determine the expected number of cycles and distribution of cycles for each variation. It was a fun way to regain my confidence with Python, and to build my own simulation. If you’re interested in reading my jupyter notebook or paper write up, let me know!
The second project was to better understand a random variable independence test of runs up and down by finding the exact distributions of number of runs up and down for small ns, and then approximate the distribution for a large n. Spoiler alert: the random package in python’s random number generator does create independent variables. Again, if you’re interested in seeing my python notebook or reading the paper, let me know!
This class was very statistics heavy. I was worried that this class would not be applicable to the analytics work that I do since I’ve rarely heard of people doing simulations at work, but the class was well worth it to better understand statistics. The first few weeks of the class is a statistics crash course, so as long as you’ve seen basic statistics before (but forget most of it), you’ll be fine.
With a 90% confidence interval, I have a 32%-44% better understanding of statistics and simulations.
5 Stars! Two thumbs up! I highly recommend this class to anyone in the program. It’s interesting, challenging, fun, and a relatively stress-free class.
MGT 8823 Data Analytics and Continuous Improvement (Spring 2021)
omsa.ga: Workload: 3.9 hours/week
This is a track elective for the Business Track. Recorded lectures were by Lee Campe, but the class instructor was James Wilburn. Despite being only 2 classes different than other tracks, the business track has the reputation for being the “easy track”, I now understand why! I spent <2 hours a week on this class. There were weekly homeworks, one project, and 0 exams. Our textbook was Moneyball by Michael Lewis. Despite the low effort course, I feel like I learned a lot about how to talk about continuous improvement.
I have already implemented learnings from this class at work. I used the DMAIC (Define, Measure, Analyze, Improve and Control) problem-solving approach that drives Lean Six Sigma to convince my stakeholders to define, measure, and analyze a process problem before jumping to improvements that they had planned. By doing those steps first, we were able to come up with better approaches to improve the process. And, by being able to articulate the Six Sigma approach, I was able to persuade my stakeholders to wait to act until after measuring, which I may not have been able to do without learnings from this course!
Our homeworks each week and our project were based on collecting data in real life and improving the processes. Because it was a pandemic, I didn’t go anywhere, so my projects were all collecting data about myself and improving my personal life including:
Is this class worth the tuition? Probably not. Am I glad I took the class? Absolutely! I read Moneyball (great book — I highly recommend it), I had fun thinking about and collecting data in my real life, I learned how to approach continuous improvement conversations, I got really good at operational definitions, and because I got an A on my final project, I now have a yellow belt in Lean Six Sigma!
3 stars! 2 thumbs up! Fun, easy class. If you’re looking for a challenging coding-based class, this is not for you.
ISYE 6414 Regression Analysis (Summer 2021)
OMSCentral: Difficulty 2.95/5; Workload 8.87 hours/week; Rating 3.37/5
Regression analysis has a polarized reputation. People fall into the following camps:
- Regression was one of the best classes in the OMSA program because it is the most applicable to my everyday work. It helped me understand the mechanics of regression, how to tune and improve regression models, and most importantly, when regression is not the right model to use. And, she gives you coding examples!
- The class was horrible. The professor was hard to understand, she makes declarations but doesn’t explain them, the multiple choice homeworks and exams are overly tricky in wording, and there isn’t enough time for the coding portion of the exam.
I think I’m in camp 1. Yes, the professor is hard to understand, but she provides transcripts of the lectures to read along to. Yes, the sometimes she doesn’t explain concepts, but the professors and the TAs are very responsive on Piazza for clarification, and there are recommended textbooks. Yes, the multiple choice questions are tricky, but I don’t think unreasonable. And yes, the coding exam can be stressful. My favorite part? She gives coding examples! These are sooo helpful for the homeworks, exams, and for future reference! My biggest complaint about this course is most of the content is covered in MGT 6203: Data Analytics for Business. This class just goes more in depth to model fit, model assumptions, and statistical inference. Additionally, the professor states statistical properties without explaining them; taking Simulation before this class was very helpful in understanding all of the concepts.
Another thing that I love about the course is content is released every 2 weeks, with due dates every 2 weeks. This gives me more flexibility and lets me either distribute the work, or plan ahead and cram everything into 1 week if I want to have more free time the other week.
The class has 4 homeworks (15% of the final grade), a midterm (40% of the final grade) and a final (45% of the final grade).
Overall, I think the pros of this class outweigh the cons. So much so, I’m going to be taking Time Series Analysis with the same professor next semester!
ISYE 6402 Time Series Analysis (Fall 2021)
omsa.ga: Workload: 15.5 hours/week
Oomph. This class was a doozy. I knew the class had a bad reputation, but because I took Regression the previous semester, I thought I would be prepared. What I realized quickly was I was already familiar with regression concepts before the course (thanks to MGT 6203 Data Analytics for Business), there are a lot of online resources for regression to help me when I get stuck, I had 3 friends to study with in the regression class, and the regression concepts are much easier because they’re a lot easier to visualize. With those components missing, Time Series Analysis was a lot more difficult than expected.
The class even started out with a letter from the professor that thanked us for taking the class. It stated that she was aware that the course had bad reviews, and has been working on revamping and improving the material. I would love to know what was improved (and how difficult the course must have been before!). But, it is clear that the professor cares a lot; She cares about her students, she cares about the material, and she cares about improving. I am hopeful that in a few years (with iterative feedback) this class can get better.
One of the hardest parts of this class is the professor does a lot of “stating” facts, rather than explaining them, proving them, or demonstrating them, which makes them hard to understand and remember. And, there are just so many new concepts to learn. I found that the suggested textbooks, googling videos, and meeting with my new study group that I found on slack helped immensely, and I would not have succeeded in this class without taking the time to find other places to learn some of the fundamental topics.
The pros of this class are exactly the same as Regression Analysis:
- She provides code examples!
- The professor and TAs are incredibly responsive on piazza if you have any questions
- The recommended textbooks are great
- Content and due dates are every 2 weeks so it allows for more flexibility of pacing content
For this class, the exams (both multiple choice and coding) were open book (ie your own notes, past homeworks, and anything on the course page but not open google) which was so helpful. However, I am a very fast test taker, and I struggled to finish the coding exams on time.
This class was a whirlwind. And I struggled. But, in the end I did get an A! And looking back, I’m so happy that I took the class. I feel like time series analysis is a skill that every analytics team is interested in to forecast performance and detect outliers, so it’s an incredibly great skill to have.
I can say that after taking this class, I can talk about the components of a time series model (trend, seasonality, and noise), the assumptions that must hold true, and how to model each. I can talk about the differences between auto-regressive (AR) models, moving average (MA) models, ARMA models, and ARIMA models, and when to use each. I can talk through how to detect and model heteroskedasticity (or variance that changes over time) and how to model it. And, I know how to create multivariate time series models (or time series models that use other time series as inputs).
I would recommend this class if you took Regression Analysis and were okay with the professor, if you don’t know anything about Time Series Analysis and think it would be a good tool to have in your belt, and if you’re not super passionate about the other statistics electives. If you have not taking regression analysis, or did not like the professor, or feel like you need to deeply understand time series analysis for work, or don’t think you’ll ever model time series, or are more passionate about other statistics electives, I would not recommend this class.
MGT 6311 Digital Marketing (Spring 2022)
omsa.ga: Workload: 3.3 hours/week
To complete the business track, I had to decide between Digital Marketing and Financial Modeling. I heard Financial Modeling was basically just an Excel class (which I was not interested in). I was also a Marketing Data Scientist at the time, so I figured that would make this class easier, and I would (hopefully) learn some new things that I could directly apply to my role.
This class was stupid easy. This is the only class that releases all of the material for the class day 1, and I think I finished the whole class in 4 weeks. I could have finished it in like 2 if I didn’t watch the lectures (but I was hopeful that the lectures might actually teach me something). I took maybe 3 pages of notes for the entire class (as opposed to most classes where I took over 3 pages of notes a week!). I’m fairly confident I could have aced the final on day 1 if I took it.
You might be thinking “Well, Riesling — you were a Marketing Data Scientist — maybe someone without marketing experience would have learned from it?” Well, I disagree for two reasons:
- You learn more about digital marketing in the marketing module in MGT 6203 Data Analytics for Business which everyone is required to take.
- One of the questions on the final (I’m paraphrasing here, but you get the point) was “Which of the following types of ads cannot be personalized? Facebook ads, website ads, or billboard ads”. I’m pretty sure anyone who has driven by a billboard and knows what the word “personalized” means could get this one right.
The only redeeming part of this course was a few of the Harvard Business Review Cases were interesting, and I did learn from those.
I do not recommend this class unless you’re just trying to finish the degree as quickly as possible. You will not learn anything.
ISYE 6748 Applied Analytics Practicum (Spring 2022)
The analytics practicum is supposed to be a place where you can apply your learnings from the program. You can do this through either:
- An employer sponsored project at your place of work
- A Georgia Tech-sponsored company with a set list of projects
I chose to do the employer sponsored project, and I worked on an email spam-bot detection project to estimate what percentage of our email click-through rate was inflated due to bots trying to confirm our emails were not spam. Unfortunately I can not share many details of my project because it was completed through my employer.
I have heard that if you do a Georgia Tech-sponsored company project, they send you descriptions of all of the projects available, you rank them, and then they assign you to a project based on your ranking and how many people want to do each project. I think some projects are done in groups. I also think that each employer has a sponsor that you can meet with to ask questions about the data and check in on the progress. Don’t hold me to any of this! This is just what I’ve heard, but since I did not complete this, it could be totally different.
The analytics practicum is maybe the least climatic way to end this program — or at least how I did. There is hardly any communication from the professor (or OMSA advisors for signing up for the class), no feedback from the teaching staff on your work, and no grades posted during the semester at all. So you’re paying for 6 credit hours (DOUBLE any other class), and getting 0 feedback or interaction.
Additionally, since I was doing an employer sponsored project, I was literally paying Georgia Tech to write a paper about work that I was already going to do anyway.
There are also a few hours of required videos to watch. They’re supposed to be from industry leaders in analytics congratulating you on your achievement, talking about the industry, and teaching you about analytics leadership. I found many of them felt like commercials for companies to work for.
Maybe if you were using the GTOMSA to transition into data and analytics, the practicum could be a great portfolio piece. For me, it felt like a waste of time and money just to check a box to get my diploma. I wish they would allow people who are already working in the field to take two additional classes instead of the practicum. I would have much preferred that! I would have actually learned something.
Final Thoughts
As I mentioned earlier, you can see my thoughts on the program holistically at my blog here: Curious about the Georgia Tech Online MS in Analytics?
If you have any questions or comments, feel free to comment them below or reach out directly.
Thanks for your time! I hope you found this helpful and informative.
