UT Austin MSDS Online Review
A review of my Master’s Degree through UT Austin.
An overview
During the COVID lockdowns, The University of Texas at Austin announced they were expanding their online Master’s in Computer Science program (MSCSO) to include a Master’s in Data Science (MSDSO). The MSDSO program of work is similar to MSCSO (along with some overlapping courses), consisting of 10 courses all offered at $1,000 each.
I was in a data analyst-type role at the time at a transportation engineering company. I also was intrigued by not having to take the GRE exam. I also already graduated from UT with my Bachelor’s in Aerospace Engineering.
Last year, UT expanded the program again adding a Master’s in AI (MSAIO).
Pros/Cons
Pros:
It’s Cheap. The $10k price tag is already quite reasonable, but the real benefit in doing an online asynchronous degree is that you don’t have to quit your job and move somewhere to go to grad school. You can keep your seniority in whatever job you have now and stay away from very poorly paid grad student work.
It’s not very risky. You should worry about the risk that you never complete the degree (and likely wouldn’t fully benefit). This risk is really significant if you invest a ton of time or money into a degree. I don’t believe the price tag or the time commitment really make this a huge risk for anyone to take. The program allows you six years to complete your course work (the MSDSO program is not even six years old yet).
No GRE requirement. Who likes taking standardized tests anyway?
Truly asynchronous. You can take this class from anywhere with an internet connection. You do not need to ever visit Austin or go to a testing “proctoring” center.
Cons:
The lack of interaction with professors and fellow students. This was something that probably sounds obvious for an online degree, but I’m not sure most people come into MSDSO with the idea that you will seldomly ever talk to your professor. I did have two final projects where I collaborated with a fellow student, but you will also very rarely talk to other students (outside of discord and the class discussion forums). Working with professors and fellow students is a pretty big part of any in-person degree, so if you come in expecting that you will be greatly disappointed.
There are few opportunities in research or applications. Related to above, since you rarely work with your professors or fellow students most will not be able to find research opportunities to work on. Very few courses offer any open-ended projects for you to work on, so if you want to apply any of the methods you learn in class you will have to do that on your own time.
No thesis track. I learned recently that the MSCSO and the new MSAIO programs both offer a thesis option for their students. I’m not sure if I would have done it if it were offered, but it sounds like an excellent way to boost your resume especially if you are trying to do a career pivot. The fact that MSDSO doesn’t offer one is really disappointing and I hope that changes soon. Related to this issue is MSAIO and MSCSO students have the option of taking many of the best MSDSO courses but the opposite is not offered for MSDSO students.
Not all of the courses are offered every semester, speedrunners beware. Some future students probably see the 10 course program of work and scoff at it and assume they can complete it easily in three (or less!) semesters. This may not be possible as not every course is offered every semester and only a small portion are offered in the summer. It’s also not announced what courses will be offered until a couple of months before the start of the semester. This might make things difficult for people trying to complete the program as fast as possible.
A word on admissions
UT admissions is infamously a black box. So anything said about what admissions is looking for is complete speculation. I would highly recommend you check the requirements and preferred qualifications and make sure you meet everything. If you do not meet one of the requirements I would go in with the assumption that I was not going to be offered admission and have a backup plan. You can also reapply every spring and fall.
The MSDSO staff have stated that there are no admission caps, so if you do meet the requirements/qualifications you will probably get in. Therefore, there really is no such thing as being a “competitive” candidate. You will also never hear back from UT why you were or were not granted admission.
Also it’s not clear the value of optional things such as letters of recommendation or GRE scores.
Courses
They currently offer 14 courses and I only took 10, so here’s my high level review of the courses I did take, in the order I took them.
Or if you prefer in tier-list format:
Probability and Simulation Based Inference for Data Science
Fall 2021
- Rating: C
- My grade: B+
This was the first course I took in the program. I do remember it being quite difficult as the probability half of the course introduced a ton of new notation for me. If you had a decent amount of stats exposure, this class would probably be a lot easier for you than it was for me. I thought it set things up okay for the rest of the program. I thought the tests were tough.
Foundations of Regression and Predictive Modeling
Spring 2022
- Rating: D
- My grade: A
This class is by far the worst in the whole program. The lectures are incredibly dry and boring. The only saving grace of this class is it is a homework-only class with no tests or projects. So, grind through it and get it done because it’s required.
Advanced Predictive Models
Summer 2022
- Rating: A
- My grade: A
This was another homework-only class. It was pretty easy and a good candidate to take over the summer. I liked the lectures in the first half of the course that covered topics that none of the other courses hit on like time-series and geospatial analysis.
Data Structures & Algorithms
Fall 2022
- Rating: S
- My grade: A
I had some prior exposure to data structures as I was working with Python daily before taking this course. This made the class a really fun challenge for me and I learned a lot. It also covered a lot of useful topics like debugging and IDEs. However, if you have zero exposure to object-oriented code and Python this class can be very tough. You are evaluated pretty much entirely on programming assignments.
Deep Learning
Spring 2023
- Rating: A
- My grade: B+
This course as since been reworked and split into two classes (they added an Advances in Deep Learning Course and nixed the final project in this course). This was probably the most difficult course for me in the program. It was mostly programming assignments were you had to set up a deep learning algorithm to solve some problem. You had to get a certain performance to get a decent grade on the assignment, so naturally this involved a ton of hyperparameter tuning. I used my own GPU to speed things up but you could also use Google Collab. The final project was a ton of work as well. The lectures are really good and you definitely learn pytorch by the end of it.
Data Exploration & Visualization
Spring 2023
- Rating: B
- My grade: A
This is a really popular class with MSDSO students partly because it is super easy, but it also has a ton of useful content in making your data visualizations understandable and clean. The whole class is programming assignments in R. The lectures are great but I just wish that more visualization options were allowed like matplotlib in Python. Also, the lack of a final project in this class is really disappointing.
Natural Language Processing
Fall 2023
- Rating: S
- My grade: A
NLP is another popular class with MSDSO students. Dr.Durrett has put a ton of work into this class. He is always adding and updating modules in the course each semester (this did not happen in any other course as far as I’m aware). The assignments are a mix of programming, a midterm, along with a final project. This is the best course in the program.
Data Science for Health Discovery and Innovation
Spring 2024
- Rating: A
- My grade: A
This was a brand new course when I took it and I really couldn’t tell. I think Dr.Parast’s lectures are the best in the program. It’s a stats-heavy course with homework and tests. It felt pretty easy but I’m not sure if the content was easy or if it was really well taught.
Principles of Machine Learning
Summer 2024
- Rating: B
- My grade: B+
The start of this course is very difficult. The first two homeworks come at you hard but after that I thought it was pretty smooth sailing. The tests were also pretty tough but generally fair. I did not like how fast and dense Dr.Klivans’s lectures were at the beginning and the textbook is also very dense and not a fun read. I would recommend the Ben-David lectures on youtube as an extra learning resource.
Reinforcement Learning
Fall 2024
- Rating: C
- My grade: A-
This was a pretty disappointing class. The lectures are very brief and are basically a summary of that section in the chapter of the textbook. The whole class is based around you reading the textbook. This might work well in-person but I don’t think this type of course works well online. It’s a homework and test class. I would recommend the David Silver lectures on youtube as an extra learning resource.
Conclusion
Generally, I’m glad I’ve completed the MSDSO program. It took me three years, but in that time I started a new job as a contract data scientist and now help lead a small team. The classes I took really haven’t provided exact “skills” that I need to do my job but really set me up with the background theoretical knowledge to know when to do something and when it’s probably not worth.
Resources
- MSDShub for many more course reviews
- MSDSO subreddit for the admissions thread to see the type of students who get admitted
- MSDSO Discord (check the subreddit sidebar for an updated link)
🤘