Data and Visual Analytics

Pros:

1. Most of your grade is based on homeworks and team project so you're rewarded for the hands-on part of the course

2. Bonus quiz opportunities are there and pretty manageable if you are on the border

3. Unlimited Gradescope submissions for HWs, so you should easily be able to get majority of the points

4. Not too difficult as someone who has full stack software engineering experience

Cons: Oh boy, there are plenty of cons!

1. Team project experience was ok. I lucked out with getting a team where everyone contributed so peer evals were easy. I wanted to do an interesting project that wasn't some cookie cutter ML model thing, but the time constraint given from the course was limiting. My team initially operated with 1 month worth of time for the project, but due to HW 4 and the annoying reports to write, there was really 2 weeks of time towards building the project. Splitting the work helps, but still we had to pare down a lot of the scope and the final product was kinda lackluster.

I recommend picking up Streamlit and the documentation is very good

2. HW 2 (the D3 one) was kinda annoying. The D3 library is very finicky to pick up and you had to match the exact HTML structure to the T in order to get the autograder points. I had scenarios where the D3 visualization was bad but passed the autograder.

3. Lack of opportunity to assess visualizations we create. For a course that's about data visualization and analytics, that piece wasn't really covered. How about add a section or multiple choice piece in gradescope where we have to write a few sentences about how an existing visualization can be improved based on the provided context

4. Professor was hardly present in the course. I'd like to see a little bit more of Polo in the picture

I’m genuinely surprised this a required course for the OMSA program. If Georgia Tech prides themselves on providing leading education, this course is certainly not living up to that standard.

The content feels like a hodgepodge of loosely connected material, nominally centered on visualization, but without much depth or cohesion. The lectures are extremely broad and don’t align well with the assignments. The workload is lighter than in most other OMSA courses. The D3 assignment in Homework 2 is the only part that takes a bit longer. D3 is tedious to learn and I highly recommend completing D3.js Essential Training by Emma Sanders on LinkedIn Learning. It will significantly help in completing the assignments.

Half of the course grade is based on a group project with at least five team members. I generally dislike group projects with people I don’t know, but ironically, this ended up being the most enjoyable part of the class.

Advice for Taking This Course:

• Take it later in your program. You’ll be better equipped to define a meaningful project once you have more context from other OMSA or OMSCS classes.
• Form a mixed group. A team with both OMSA and OMSCS students will bring strengths from the analytical and the computer science areas.
• Choose a manageable project. Pick something with a clearly defined goal and a dataset that can be obtained. Start with a simple idea - you can always add complexity later.
• Find your group early. Be proactive. Try forming your group at the very start of the course (or even just before it begins).
• Meet weekly and stay organized. Set an agenda, assign roles and tasks, and ensure at least one team member keeps the project on track.
• Check the checklists! The course, and especially the project, is full of detailed “to-do” lists that can make you feel micro-managed to the n-th degree! If you want a good grade, be sure to complete every item on those lists.

I was completely swamped with my job when I took this course and I therefore didn't put much effort into it at all, and I did not do well in it because I refused to cheat to complete the assignments. That being said, it would have been really easy to cheat and complete the assignments correctly to get an easy A. Beyond the grade, though, the course is beyond awful, and the funny thing is that everyone knows it but nothing has been done about it. There's nothing much I can add beyond what people have already written in this forum and what your classmates in your group are probably saying about it. This course needs a revamp so badly that it honestly should be the first revision made to the OMSA curriculum once the powers that be finally get to it

This class is nothing short of a disaster. On paper, this could be a blockbuster class in the program. Visualization is such a critical piece in the field of data science. Telling a story with data is a cornerstone skill in the profession. Often times, its more important than the analysis itself. Conveying insights can be so challenging with numbers. Visuals can piece layers of complex understanding with imagery that sells an analysis on its own. Some of the most compelling projects in my career have been propelled because an executive remembered a visual we created.

This class promotes none of those critical skills. Lets step through what the semester looked like:

1. Lectures: The lectures had no relevance to the class. They were often vague, scattered, and completely disjoint from the class itself. You could get through the class with an A without tuning into a single lecture. That is not learning. I'm sure Polo is a qualified professor, but his delivery in these lectures was disappointing.

2. Teaching Assistants: I did not engage personally with the TAs, but I observed many interactions with the TAs on the Ed forums. Quite frankly, many of the TAs are extremely arrogant and condescending. There was some confusion around technology in my section, and the TA's response was 'read the syllabus/instructions' and 'its not our problem if you can't find the answer'. The questions being asked had nothing to do with either of those things. Additionally, one of the technologies required to be used went down for a period of time. It was required for the homework. Rather than allowing a grace period for the downtime, the lead TA simply said 'not his problem' and many students were left with a 0 for that portion. For this class to get better, this has to stop. I can go on and on with more examples of this type of behavior. If this slate of TAs is unable to change, then they need to be replaced.

3. Content: There was very little content relevant to the world of data visualization in this class. The first assignment was some strange actor graph chart coupled with some other obscure analytics that had no relevance to visualization. The second was the infamous D3 assignment that is irrelevant. I think half the class just didn't do it because its not necessary for an A. The third assignment was doing the same analyses in like 5 different data platforms. The fourth was coding up a random forest from scratch. If you had trouble at all, these are all things that a Saturday morning, ChatGPT, and cup of coffee could figure out before lunch time.

The project was a total joke. In our peer grading, its very obvious that very few groups took this seriously. One team submitted what was very obviously a ChatGPT-designed poster. The project is worth 50% of your grade, to see so little effort was disappointing. The design of the project was also confusing. Why the emphasis on research literature? We a 6-person group, which meant we needed to submit 18 peer reviewed articles summarizing all of them + complete a project proposal within a 2 page limit. Thats just frankly unacceptable. Its not possible to generate a robust proposal when Im given less than a page to work with after the literature survey is complete. There was also an idyllic obsession with 'novelty' in the project that is just frankly absurd. Who cares? The point of this class is to develop skills, conduct complex analysis, and be able to show it in a way that a casual observer can understand. The focus on 'finding the next best academic thing to improve on' just propagates the idea that this profession is out of reach of the most casual observer. Its not.

None of these things had any relevance to data visualization. Which chart should I use for what purpose? How do I use color shading to draw attention to important things? What are the best ways to build a presentation to tell a story about data? How do I layout a slide in a presentation for maximum impact? These are things you'd expect out of a data visualization class and they just simply are not taught here. Why are we coding random forests in a visualization class?

You'll have to take this class, so there's no getting around. Start homework early (you get ~ 3 weeks), make sure you have a decent project group, and just recognize its going to be a slog.

I took this class alongside the Deep Learning course, and the difference was very noticeable. Deep Learning introduced a new homework this semester (Spring 2025), and most of the assignments felt fresh and up-to-date. In comparison, this course feels stuck in 2017. The lecture videos, homework, and tools all seem outdated.

Deep Learning has an enrollment of about 150 students and the professor holds regular weekly office hours. In this course, with over 1,000 students, the professor is MIA until the very end when he asks for reviews. The whole experience felt more like a generic online course. The TA office hours are text-based, which makes the course feel even more impersonal.

I had high hopes for this class. I took the recommended Data Visualization courses by Curran Kelleher and was excited to make creative visualizations and learn about design. Instead, most of the work involved recreating outdated D3 charts. There's little focus on good design. Only one assignment asks you to apply basic design ideas to a table, and it's worth just five points. The rest is just copying and submitting through GS.

Many students just rely on GPT to complete the assignments and project. Some even brag about not learning D3 at all and just using gpt or cursor. The first homework was basic SQL and Python, similar to LeetCode Easy problems. If you couldn’t solve them, GPT could do it for you without any issues. I was also surprised by the low level of questions on the forum—some students didn’t even understand object-oriented programming.

The final project was a total letdown. I thought we’d be making something informative and creative like a NYT or Economist-style visualization. Instead, it was just a checklist. As long as you answered the Hellmeier questions, you could turn in the ugliest chart imaginable and still get full credit.

Group work was a disaster. People didn’t know how to use Git, pushed API keys, uploaded giant CSVs and parquet files, and even dumped raw ChatGPT output (emojis, instructions, everything ) straight into the report. They didn't even bother to check the GPT output, like how do these people get in and how are they allowed to graduate?

I did hear of people who got their grade reduced for not contributing to the final project though, so that was a redeeming quality.

Homework Summary:

Homework 1 and 2: Basic SQL and Python, some D3 and Tableau. Mostly just copying old visualizations.

Homework 3: Simple data cleaning with PySpark and Scala. Claims to teach Docker and cloud platforms like AWS and Azure, but in reality, you just follow instructions in a pdf to make an account and complete exercises in a jupyter notebook. Add it to iCDA?

Homework 4: Basic machine learning with sklearn and some algorithms from scratch. It was a good assignment but felt out of place in a data visualization class. Move it to CDA instead?

Suggestions for Improvement:

• Add a short proctored design quiz that helps students recognize good and bad visualizations.

• Include a homework on deploying visualizations so students can share their work with classmates

• Replace the final poster with the actual deployed visualization. The poster and stringent rubric really made me feel like I was in middle school.

• Include a simple proctored assignment using Tableau or D3 to confirm people can actually program.

• Emphasize teaching design principles so students can create clear, effective visualizations.

Overall, this class felt like a massive waste of time.

Good things:

• Low workload if you have a good group and can code reasonably well

• Don't have to watch lectures because they're pointless

Bad things:

• Pointless lectures, don't relate to the assignments or the group project at all and contain high-level information that you could learn from 5 minutes of Googling

• If you have a bad group, get ready to carry. You will get some slackers guaranteed, happens in every group project in OMSA/OMSCS

• Very pedantic project requirements; lots of time spent writing silly reports

• Assignments are not useful at all and barely scratch the surface of what could be considered useful information. Example: go make an account on AWS, then do this very basic python coding. Ok, now go do the same thing on GCP! Sorry, but this sucks.

A piece of advice: make sure you get at least one OMSCS student in your group, if not more. You have to make an interactive app as part of the group project and you don't want to be stuck with only people who have never done full-stack development (at least as a hobby) trying to code your app up. Also, brush up on git! It will help a lot.

I really wanted to like this course as it covers some really important topics in the world of data science which could really set this program apart from others, but I feel it fell short. Your experience in this course may also be heavily influenced by the group you end up with for the project so make sure to get on top of that right when the course begins.

Some of the material covered was awesome, such as data collection, cleaning, and integration as well as sql and scaleable computing. The course also covers data visualization which again could be a very valuable topic, however, instead of things used in the industry such as tableau, power BI, and python visualization packages, you go in depth on D3 which just felt like a huge waste of time. Additionally, the last few weeks of the course along with the last homework covers modeling topics that are taught in more depth in other courses, such as tree based models and clustering, so I didn't really understand the point of those. I felt this time could've better been used covering the more relevant data and visualization topics I mentioned previously.

Another issue is the lectures don't add value in this class and can be skipped aside from the bonus quizzes. They're very high level, only provide a short overview of the topics, and aren't connected to the assignments we're graded on. I do value that the class forces you to learn code on the fly as that's similar to what you experience in the real world, but when it comes to education, applying and reinforcing the lecture material in some way is the best way to retain and learn information, and this class falls short on that.

As for the assignments, they're definitely long and the difficulty of them depends on your background. Homeworks 1,3, and 4 are more python/sql based so how long it takes you definitely depends on your level of python proficiency. If CSE 6040 was easy for you, you probably won't struggle too much on these so YMMV here. On the other hand, homework 2, the D3 assignment was the nightmare it was expected to be, and the fact that most of us will probably never need D3 makes this even more frustrating. As others have mentioned, you can score very low on this assignment and still get an A in the course, so that may be the move for you if you don't want to devote a crazy amount of time to it. I also found coding the random forest from scratch in homework 4 difficult, but if you're comfortable with OOP and have taken ISYE 6740, you may not find it challenging.

Lastly, there's the project worth half your grade. As others have mentioned, it's graded pretty leniently and your experience on it really depends on how good your group is. Generally, the people who'll be the best teammates are the ones who are actively looking for groups on the first day of the course, so I recommend being active on ED from the get go to find people. If you follow all the guidelines, it's graded pretty leniently so you pretty much get out of this project what you put in. You can be ambitious and end up with a project that you can advertise in job interviews, or you can do something small and still get an A. Again, YMMV depending on the group you end up with and how complex of a project you want to go for. Just be prepared to work on it throughout the semester to meet the deadlines instead of cramming, and if you end up with a bad group, be prepared to do a lot of heavy lifting. I'm honestly mixed on how worthwhile I think the project is. On the one hand, it gives you the opportunity to really push yourself and apply some of the material covered in this class, but on the flip side, the guidelines are so open ended that you may also not get a whole lot out of it and may have to deal with some online group project chaos.

This class is so close to being great if it just makes some tweaks on how it operates, but despite how challenging and time consuming it may be for many, it's also not difficult to get an A due to the unlimited assignment submissions, lenient project grading, and bonus quiz opportunities.

This class was probably a lot harder before chatGPT. -Autograder still sucks though -Chat can get you 50% on HW's, a little leg work can get you to a 65-75 -That's all you need because half your grade is the group project.

If you have a strong project group, don't kill yourself over getting 90+ on the homework.

Overall I'd say it's pretty easy to get a B, and pretty hard to get an A. I ended up with an A despite getting D's on two or 3 of the HW's. Lectures are useless, you and chatGPT teach yourself everything.

How to succeed: Know javscript for D3. Know SQL. Know PySpark. Then you will be good to go!

This class is not really related to visualization. Rather, they put together a bunch of technology that is relevant for data engineering/high end data scientists.

If you want to get something out of this class, when they are doing AWS/GCP/Databricks/SQL, with the extra time that you have study those 4 in deeeeep depth. Those 4 will be very relevant after you graduate. Other things they put emphasis on such as D3, final HW which is little bit of classical machine learning and a final project, DO NOT WASTE TIME ON THOSE. For classical machine learning, just take computational data analytics class (it is an awesome class BTW).

Anyways, Overall this class focuses on all the wrong things and flies over the important ones. I would just spend this semester looking at the above 4 things and ignore other things. Yes you might end up with a C, but you will be much more marketable than someone who got an A and knows how to code D3.