Course Details

Pros: 1. Easy A? 2. Honestly struggling to come up with positives 3. Cons: 1. Most of the first half is useless 2. Meaningless calculations to be performed by long hand 3. False precision on the Quizzes will ding you Detailed Review: I'll set aside the garbage experience with peerceptive - others have addressed this. My issue with the class is there's nothing about "predictive modeling" to be learned here. The first half has some Time Series (1 module) and that too is very limited - they spent a lot of time in some spatial modeling which probably 5% of the world uses. The first half half is almost entirely useless. The second half has a lot of matrix math calculations to be done by long hand, SVD, PCA etc. which is covered elsewhere in the program. This course is highly disappointing and waste of your time. Really they should scrap this course entirely and redesign it to truly teach predicitive modeling.

Pros: 1. Cool material, some of which you see nowhere else 2. Easy A 3. Professors are actually alive on Ed Cons: 1. Lots of self-study needed to understand everything 2. Horrifically ran course this semester 3. Unclear expectations on assignments Detailed Review: Okay, first, let's talk about the horrific way this class was run this semester. So, peerceptiv is garbage software with a ton of issues. That isn't the class' fault, but ML had figured out workarounds with peerceptiv long before APM did. There were also not anywhere near enough TAs. The one or two office hours I did go to were well done, but the Ed board was pretty dead the entire time. Most of the time, we would ask the same questions over and over again and they remained unanswered for over a week. Just painful. There are two HW assignments most weeks: a peer-graded one and a staff-graded one. You get a rubric to grade peers for the peer graded one which has solutions. The staff-graded HW has no solutions though, so good luck figuring out what you did wrong. The turn around time has been absolutely atrocious. I'm finished with the course now and I just got back the HW grades from 3 weeks ago. A lot of the time, the grading is off since expectations weren't clearly outlined and the rubric expected more than the problem statement said. Regrades have also taken an eternity to be done. If you ask for clarification on what you got wrong on staff-graded HW, good luck on getting an answer back anytime soon. I think the TAs are just completely overworked. Grading is all HWs and little mini quizzes every week. Some of the quizzes are some BS but overall, it should be very doable to maintain a 94+ average with them. The HWs, with some regrading, should also be easy to maintain 94+ with. Your lowest entire week (so quizzes + any HW) is dropped at the end too which makes it even easier on the grade. As for the material and teaching, the first half of the course to me just felt like reading formulas off of a slide. Very little intuition on why things worked and a lot of "plug and chug". It was a lot of MVN, applications such as geospatial models, and time series. So, a lot of understanding the material was self study. Unfortunately, the book provided for this half of the course wasn't too great either in my opinion. Would recommend reading https://link.springer.com/book/10.1007/978-3-319-29854-2 (free with UT login) instead. Much clearer. The second half was much better in my opinion when it came to lectures and work. It was mostly about like matrix methods. Sure, we've seen SVD/PCA a gazillion times before but there's a lot of cool new methods like spectral clustering and a lot of blockmodel stuff that was great to learn. One plus is that the professors actually chatted on Ed every once in a while. That is a huge plus since that isn't really the norm in this program. Maybe, having two professors in a class helps to free up time for them to be active in a forum.

I'm not trying to be unkind here, but these lectures are just awful. Lectures were comprised of just super, super slow talking, and walking though poorly written and hard to read equations. I'm three weeks in and already in "I can't believe I actually paid for this" territory. They also moved the course to Canvas this semester and, frankly, they don't seem to really know how to use it. Aren't there learning technologists to help course staff with this sort of stuff (rhetorical question - there are). Canvas is just super disorganized, peer grading is super disorganized (I have to rate this person on their feedback, but can't see what grade they gave me? What?) If this is the part of the course that folks think is well done, I cringe thinking about what the second half will be like. They've also set this up so that you can't access any of the material for the next week's assignment (including lectures) until Thursday of the current week, and all the assignments for the next week are due the following Friday at midnight. So you basically have one weekend to watch all of the lectures and complete the work per module. Not very accommodating to folks who work full time, but I suspect that may be the point. Would be remiss if I didn't point out that both the lectures and the slides (and in one instance, the sample code) are absolutely riddled with errors, with scant attention seemingly given to correcting any of this in a really meaningful way (some corrections scrawled underneath some of the lecture videos in a few instances, but that's about it). The assignments for the first half of the course really just feel like pointless busywork. I also have to call BS on the other reviews here. Not sure which course they're talking about, but it isn't the one we're in this semester. Super disappointing overall. If this is the future of online education, count me out. ______________________________________________________________________________________________________________________________ UPDATE: Heading into week 5 of what is effectively a 10 week course (2 / 12 modules are "optional") and we've received, wait for it...grades for only the first assignment. That means that we'll effectively have completed 80% of the deliverables in the first half of the course while only having gotten feedback on a single deliverable (or 20% of the deliverables for the first half of the course). Not too great, if you ask me. Also, they forgot to release the week 5 material until prompted by students and there was basically radio silence (i.e., not a single question answered by course staff) in edStem for maybe 7 - 10 days (again, until a student piped up and complained about this). Instructor response was basically a polished version of "answer your own questions". Really, here it is verbatim: "I have been reviewing the activity on Ed Discussion this week and have generally been pleased with students helping each other interpret homework questions. In particular, figuring out what is meant by a bounding box means really understanding what makes something a point pattern. I have posted an answer now, but would like to take this opportunity to share my appreciation to the students who have responded to these posts. This is exactly the type of discussion that helps learning and how the Ed Discussion tool should be used. As for the activity of instructional staff, we are doing our best to keep up with the volume of administrative issues that have come up this semester. We monitor Ed Discussion routinely, but cannot write a response to every post." When you are actually in the course, you will understand the context for the question that prompted this response, and why "figuring out what is meant by a bounding box" *isn't* just a matter of "really understanding what makes something a point pattern". ______________________________________________________________________________________________________________________________ Week 5: They "cancelled" the staff graded HW for week 5, ostensibly to give themselves more time to grade (if only there were some way to predict - in a course on advanced predictive models - how much effort would be required to manually grade these assignments). Instead, we have some clearly very hastily thrown together peer graded homework which amounts to a bunch of pointless busywork. ______________________________________________________________________________________________________________________________ UPDATE 2: Not sure what's going on with the prior reviewers, but I am personally finding the second half of the course much better put together than the first half. Yes, it's true - there are no coding walkthroughs. But it's Python...Google it (or already know it). Lectures are much better, instructor is much, much more engaging. Why we couldn't have Professor Sarkar teach the entire course, I have no idea. I have little doubt that this would be a much less demoralizing experience if she did.

Pros: 1. Material in first half was good 2. Not a huge time commitment Cons: 1. Material in second half was not useful 2. Quizzes and staff graded homework grading annoying at times Detailed Review: The course is divided into two separate sections. The first half covers the theory and practice of modeling using time series and spatial analysis. The second half goes over matrix completion and network analysis methods. The first part was useful as most people will need to deal with either time series or spatial analysis at some point in their lives, if not both. My only complaint was that the material seemed rushed since it was only half a semester. It would have been nice to go over the finer points of the theory and discussion of other topics. Most of the homework's involved doing some analysis of a dataset using prebuilt R library packages to perform. Although I did like this, I do wish they could have covered a bit more on the theory and algorithm side as well. You are also forced to do the assignments in R during this portion, which isn't ideal if you work day to day in python. Second half spent several weeks going over linear algebra topics and dimensionality reduction that is already covered extensively in other courses in the program. These weeks honestly just felt like a waste of time. Only silver lining was that these assignments were extremely easy, and I spent less than 3 hours per week on the course during time. The last couple weeks were then a massive info dump over network analysis. Just skimming through the slides, most of the methods covered have fallen out of favor to more modern methods like graph neural networks, which isn't covered. Needless to say, wasn't a huge fan of this portion of the class. The grading on the quizzes (and yes they are quizzes, not "practice" as they are labeled) and homework's were annoying at times. On some of the quizzes, the rounding policy for the fill in the numerical answer questions seemed inconsistent. Sometimes the solution rounded their answer, and sometimes it didn't. This made trying to determine what to input as the last digit of your answer in order to receive credit an unnecessary chore. Then for the staff graded portions, they just gave you a credit / no credit for every portion. If you made a tiny error, it could end up costing you much of the assignment. If they are going to grade like this, I didn't see the point of having a human review it since there is no partial credit for having the correct approach. They could have just converted this into a fill in the blank / multiple choice like every other course in the program honestly. Overall, this was probably my least favorite course in the program. If it only covered the first half, but in more detailed, my opinion would likely be different.

Pros: 1. Very useful material: covers many relevant spatial, temporal, and matrix applications 2. Assignments were fairly useful and consistent with the weekly topics 3. Workload was manageable in the summer listing, especially compared to past iterations Cons: 1. Second professor has fairly poor lectures 2. 3. Detailed Review: This was course 7 out of 10 for me in the program and I have to say that I quite enjoyed it. The first portion is R based with geospatial and time-based analysis. The topics explained were very relevant, although they were taught at such a rapid pace that it was hard to do a deep dive into the concepts before the next big group of models was introduced. I feel like this could have been a whole course in itself, but it makes sense why everything in the course was taught together. The second portion is Python based with matrix-based models. While I feel that this portion was even more helpful/relevant for general data science, the lectures were not taught particularly well. Overall this course was quite good and enjoyable. The evaluations were very fair and the workload for the summer was very manageable. This was one of the better organized and designed courses in the program.

Pros: 1. Regular weekly homeworks and no exams 2. Assignments focus heavily on application of the concepts in lecture 3. Covers a broad area of topics Cons: 1. Difficult to work ahead due to material release schedule and office hours 2. Expectations on homework are sometimes unclear, no rubric is provided for staff graded homework 3. Shift in teaching styles between first and second half of the course can be jarring Detailed Review: This course has a pretty simple structure of lectures released weekly accompanied by graded practice problems, peer graded, and staff graded homework assignments. 10% of the grade is for participation and peer grading properly. Honestly they should just remove the participation element, as its presence tended to decrease the quality of discussion on Piazza (the TA explicitly called out people making minor edits for "participation"). It appears that based on feedback from previous semesters, weeks 6 and 12 of the material were made optional so that the 10 weeks of the summer semester had 10 weeks of content which was greatly appreciated. The first half of the course spends half the time on time series data and the other half on spatial data. The professor's lectures are easy to understand and supplemental materials are provided if needed (I found them to often be beyond the scope of the lectures). There are also separate lectures for coding demos, which make starting the homeworks much easier as they introduce you to working with the often niche packages necessary to complete the assignment. As a result, previous R knowledge is not necessarily required but recommended. The second half of the course focuses on modeling with matrices. The lectures overall were a pretty gentle introduction to the linear algebra concepts necessary for later on, but incomplete code snippets in the lectures or visuals with no code as context at all made starting the assignments more difficult. The second half of the course requires Python. It is expected that you have a grasp on the concepts of numpy such as indexing and vectorization, as these are necessary for the assignments to run in a timely manner (or at all), but not explicitly taught or even really mentioned anywhere in the course. Unlike the first half of the course, there are less supplemental materials provided. A pain point I personally encountered on the spectral clustering portion of the class is from differences in the implementation of this algorithm. The lectures are based on a specific paper, and if you don't use that implementation specifically, the hoemwork will not come out properly I found. There were code snippets at least provided for most of spectral clustering, so just know that using those for all the assignments regarding spectral clustering is best practice if you want your answers to come out matching the expected answers as best as possible. Almost every week there is a peer-graded homework and a staff-graded homework. The rubric for peer grading is always the same on a 4 point scale with relatively subjective criteria determining the grade. What constitutes as "minor errors" is poorly defined, but luckily that means in most cases this rubric will work in your favor as people generally lean towards giving higher grades with this rubric (the average grades on all peer graded assignments were around 3.5 or higher). The staff-graded homework is always on a 20 point scale, but no rubric or answer key is provided and feedback from the grader tends to be vague, so attention to detail on these assignments is important to not lose a significant number of points. This course is pretty well structured all around and covers a broad area of topics to the extent that those who wish to dive into them deeper will have the right toolset to get started. The lectures, practice, and other provided materials will set you up for success in this course.

Pros: 1. No exams. 2. Time commitment was consistent throughout the semester 3. Interesting geospatial section Cons: 1. Classes and assignments are released on an awkward schedule so you can't plan ahead well 2. Second half is definitely worse than the first. 3. Expectations on homework was unclear at times Detailed Review: This course had some ups and downs. The material in the first half was easy enough to follow along. It's taught out of the math department and like most of those- you leave the class with a bunch of theory and not a lot of strategy of how you would apply what you learned to real-life situations. Why would you select some of the algorithms? No idea. I can tell you what the algorithm does but not why you would apply it. It would be good to review some linear algebra ahead of this course- specifically eigenvalues and eigenvectors. Most of the math is done in R or Python but there is one assignment that has some calculations by hand you have to scan in. The Data Viz class gives you enough of an R foundation to feel comfortable here- so much so that I'd consider it a pre-req. There was one library we had to use called likifit that had some of the worst documentation I've ever seen and seemed to only be used by the package authors so there was very little available to learn from. The first half of the course is well taught and you feel like you have a good start with the code shown in lecture to figure out the homework problems. By the second half, there is little to no review of the python required to manipulate the matrices- just the math formulas and even those are difficult to follow the what/Why behind them. The lectures had to be watched a couple of times and it was hard to find supplementary material on some of the topics outside of published papers- which is concerning as that means it's not used outside of academia. That translates (at least to me) as likely not useful for application to real-world datasets. You are expected to know Python fairly well, especially with regard to matrix manipulation, but there are no restrictions as to what packages you use so if you are comfortable with numpy- you will probably be ok. Students were good on Piazza about sharing youtube videos or other resources that better explained certain topics. The TA's seemed overtaxed and were completely absent from Piazza for days at a time with questions about clarifying homework questions left unanswered. There were office hour sessions but they were sometimes cancelled with no notice and it was difficult to know if they would be there at the time they said. The course material is released on an awkward schedule so sometimes lectures will release at the beginning of the week and sometimes at the end- with homework due a week after the lecture release. It made it difficult to plan ahead if you had a set schedule of when you had time to work on the course. The grading is done half by peers and half by the TA's. The grading for peers is on a 4 point scale (usually) which was generally insufficient. Some peers would mark off an answer with a whole point (equaling 25% of a grade) for missing something small or being off by a tenth of a point. The questions would ask you to discuss certain points and for this you generally need a crystal ball. They will be looking for something really specific but the question is something like "discuss the differences in the two matrices" or something, No matter how in-depth your response is you will get a point deduction because you didn't mention "density" or something. They do not release any solutions for the staff-graded problems and the solution keys released for the peer graded ones contain no code- so there's no way to see what an optimal solution would look like. you have to grade 5 peers for each assignment and they are not always trivial to grade so you should expect that to take some time- especially if you want to give any feedback at all. The practice problems are an annoyance. They are required to do but you get exactly one shot at answering them correctly. For some, you may have to answer all 5 parts before hitting submit and if you got the first part wrong, it cascades down and tanks the weekly grade. I wish they allowed for multiple attempts- it is practice after all. The latter part of the course had practice problems that required non-trivial code to be written to answer the questions. Sometimes this would help you jumpstart the homework but other times it was a lot of work for just a couple of points. For the second half of the course, you will definitely learn the material through the homework assignments and not so much through the lectures. On grading- there was a very generous curve applied. I think you could miss one peer grading without penalty, they dropped the lowest 2 weekly question scores as well as the lowest peer grade and the lowest staff-graded homework. Bonus points were based on the peer reviews you made but mostly based on piazza contributions so make sure your name shows up in replies or questions several times in the semester to make sure you don't miss out on that grade bump.

Pros: 1. Interesting Material 2. Practical 3. Broad range of material Cons: 1. Brittle grading rubrics 2. Bonus points made Piazza very annoying 3. Surprise bonus points announced at the end for attending office hours. Detailed Review: A few of the classes here have bonus points for participation on piazza. As usual, this made for - a series of vapid posts like "here's 10 pandas functions you can use" and "here's a link to some random blog post". - a race to answer student questions whether you have the answer right or not. The TA's endorse the right answer, so you have a 50/50 shot of getting endorsed this way. The grading rubrics were a bit weird, sometimes with arbitrary ways to lose points. For example, I lost points on one question because I had the title of my plots on the bottom, following the style of rmarkdown, rather than following the rubric's requirement that they be on top. The material itself was interesting though, and the assignments themselves were practical. They covered a pretty broad range of material, and I was surprised how much depth they managed to get into. I feel like I actually know what I'm doing (or at least I'm well on the way to figuring out the details) in time-series, geospatial, and network analysis. I've applied some of the material in work already. I wish regression was a bit more practical like this class was.

Pros: 1. No tests/quizzes 2. Good starter code in the first half 3. Pretty useful if you work with time series or geospatial data 4. Preston the TA was active on Piazza and was pretty open about things Cons: 1. Kinda crammed into the summer semester, twice during the semester there are two homeworks due within one work week. (monday, friday) 2. Sometimes get docked points for unclear expectations on homework 3. Sometimes R packages/Python Libraries get mixed up and ends up being a waste of time Detailed Review: The first half of the course is in R, the second half is in Python. I didn't know much R before taking this course. Some students said that taking data visualization before or during this course would be helpful, but I got the hang of it without viz. Dr. Calder (first half) provides some clear and documented starter code and usually has a lecture devoted to walking you through it. Her topics cover some time series analysis and geospatial which was really interesting to me as my job is pretty much entirely entirely that, so I was able to apply it outside of the class immediately. Dr. Sarkar's half of the course is more about PCA and matrices, which is just a tad too removed from practicality for me. She doesn't provide any starter code, so knowing some NumPy and Pandas is a big leg up on this section. The assignments are challenging but pretty interesting and are mostly code-based. I'm pretty confident that almost everyone passed the course. Getting the right version of the required R packages for each assignment always ended up being a constant annoyance for me. Wish they'd provide a requirements file for all the packages and version numbers needed to do the assignments. Reading the documentation on these packages isn't stated but should be required for understanding all of the parameters that go into particular functions. Taking this course during the summer was doable but they certainly didn't cut anything out. There are 12 assignments and 10 weeks so, you end up having two weeks where you double up. There's no way to work ahead if you're traveling or really busy during those weeks so it can be hard to plan around. On those weeks, one is due on Monday and the other is due on a Friday, which is pretty awful if you work a demanding schedule. Preston, the TA, allows for a one-time extension if you contact him. The homeworks are all peer-graded and the rubric is given to you is very detailed. You also get evaluated based on how well you grade homeworks. Doing 5 peer evaluations and following the rubric took at least an hour per assignment. Some open-ended questions wanted really particular answers that were pretty tough to get without reading the minds of the instructors. You do end up getting bonus points for Piazza participation and doing well on evaluations to make up for those, though.

Pros: 1. Weekly HW assignments take some effort but are very doable 2. Material (especially in the first half) is interesting, uses geospatial data and real life time series data sets 3. TA is extremely active on Piazza, both professors seem like nice people Cons: 1. Professors switch halfway through the course and the second half instruction is worse than the first 2. Second half of course lectures are unclear sometimes 3. Detailed Review: This course left an overall positive impression on me. Grade is made up of weekly homework assignments which can take from 2-6 hours, but are very doable/satisfying when you finally crack them. There are some straight up math problems that you'll need to solve on paper/draw on a touchscreen and submit. The first half of the course is taught by Professor Calder and covers time series and geospatial data. She uses R and provides helpful starter code and code walkthrough lectures. The material/math can seem a bit intimidating, but Professor Calder did a good job of making it intelligible. The second half of the course is taught by Professor Sarkar, and deals mostly with matrices/PCA/clustering. Professor Sarkar uses Python, and does not provide starter code (nor does she copy the code into the lectures, but rather includes screenshots of code snippets, which you can type out yourself). Professor Sarkar seems like a very nice person and I could tell she was trying to introduce us gently into matrices/linear algebra, but I found her lectures very hard to follow, especially as the course went on. She includes a bunch of graphs but never puts any axes labels on them so most of the time I had no idea what she was referring to. By the end of the course I ended up just watching her lectures on 2.5 speed to get through them in the off chance that she said something useful for doing the homework, and always googled how to do the homework. Preston the TA is amazing; he's the most responsive TA I've seen on Piazza. It could be because it was a small course so he could be more responsive to everyone's posts. The homework is peer graded and I felt my peer graders were fair. One thing, make sure you include a title, axes labels, and figure description on all your plots/graphs! That was always on the answer key/grading rubrics we used. First half professor/lecture quality - 6/7, second half - 3/7.