I finished with a 100%, but realistically it felt more like I deserved a 60%. This course is extremely challenging—by far the hardest class I've taken in the entire OMSA/OMSCS program. It’s the first time in OMSCS I felt like I was genuinely doing graduate-level math and facing graduate-level demands. Each module required roughly a hundred pages of dense textbook reading (ESL—not ISL) and/or multiple research papers.

Lectures themselves weren't bad, but mostly focused on mathematical derivations of various algorithms. Unlike CDA or ML, where you can rely on lectures recorded at other schools like Cornell, MIT, Stanford, or even YouTube (StatsQuest, Ritvik Math), the algorithms in this course are pretty obscure (things like tensor decomposition, B-splines, SVT) so you're often stuck with just the provided materials and maybe some obscure online resource from a lesser-known university. ChatGPT was probably one of the most useful resources for clarifying these concepts.

Assignments are pretty tough, although grading is extremely lenient (probably as lenient as CDA but without bonus points). You get some skeleton code separately provided, which helps tremendously, and there have been one or two questions where I just couldn't understand the material but was able to just reuse the example code and still get full marks. Without it, homework difficulty would easily hit 10 out of 5. Overall, assignments vary quite a bit in difficulty, but that probably depends on your background as well. For example HW2 is pretty easy if you've taken DL/CV. Exams are significantly harder and more time-consuming than homework assignments, and unfortunately feedback on your work is basically nonexistent.

The instructor is completely MIA. However, one thing that stood out positively was that all TAs were actual PhD students—something rare in OMSCS. Unlike some courses where you can tell students know more than the TAs, here the TAs clearly knew their stuff. Office hours were pretty just quick Q&A sessions, nothing instructional or proactive like in DL. On the upside, TA responses on ted were fantastic. The smartest students in the program seemed drawn to this class, so the forums had unusually high-quality discussions. We had one student, Joey, who basically rewrote and updated large portions of example code, which was extremely helpful.

If you found CDA remotely challenging, avoid this class. For students who thought math in introductory courses (like SIM) was rigorous—definitely skip this one. To realistically prep for this class, Gilbert Strang’s linear algebra lectures on MIT OCW or Steve Brunton’s SVD playlist on YouTube are good starting points.

TLDR The class is genuinely very tough, the instructor is absent, and the grading is absurdly forgiving. Still, this was definitely one of my favorite courses in OMSA/OMSCS because it felt like authentic graduate-level learning. Highly recommended, but only if your math skills are already very solid.

This course covers a wide range of topics related to machine learning, optimization, data preparation, dimensionality reduction, and image manipulation and analysis. I was able to earn an A, but compared to other courses rated this high, the overall experience, especially in organization and course management, was somewhat disappointing.

In terms of difficulty and time requirements the course is on par with ISYE 6740/CDA. The course has a similar structure in terms of lectures and assignments.

Pros: • The material and assignments are interesting, engaging, and rewarding once completed. • Grading is fairly lenient, and you still receive a good amount of credit even if your solutions are incorrect or only partially completed.

Cons: • The professor was completely absent from the course beyond the lecture videos. • Office hours were not particularly helpful and limited to answering direct questions. • TA responses on the forums were often brief and sometimes came across as curt or even rude. Questions took a long time to get responses, and some were never answered. I also received conflicting answers to problems from different TAs. • Lecture materials contained minor errors and appeared to have not been updated in some time. The provided code files were not well maintained. Corrections, when made, were not clearly communicated to the class. • While the lectures were concise, some lacked depth and would have benefited from more thorough explanations. • Grading took forever, most of the assignments including the first exam were not graded prior to the drop date. Advice: • Learn and get comfortable with R. While the lectures use MATLAB, sample code is also provided in Python and R. Although I’m very comfortable with Python, I found about half of the assignments easier and cleaner to implement in R. • Form a study group. Given the limited support from office hours and forums, you’ll need to rely on outside resources like YouTube or collaborate with classmates to understand the material. • Participate in the forum to answer other students’ questions.
• Start homework early! The assignments may seem daunting at first and you might not even know where to begin but stick with it. Sometimes it took me a day or two before things clicked. Despite my frustrations, I still recommend this course. The content is solid, and the assignments are genuinely rewarding learning experiences. I just hope the course gets a bit of a revamp in how it's managed—and it wouldn’t hurt for the professor to make a cameo.

This is my 7th class taken in OMSA (taken iAM, iCDA, DAB, SIM, CDA, and Digital Marketing prior) and most of my experience with Python/R comes from this program. This was by far the hardest class I've taken and I'm definitely glad its over with.

Most of the content is really interesting especially the image manipulation assignments, but if you haven't taken DO like me then the optimization modules might end up killing you. You get two weeks per homework so that about a month of time spent working on the optimization homeworks were the most intense it got. But given that we have two weeks to work on everything I still don't think I ever got to more than 20 hours of work for this class in one week. Some assignments I could even do in the week we got so that really brings down the average time spent per week.

The sample code comes in Matlab, Python, and R and the slides/videos all use Matlab. I feel like in some cases it was more convenient to use R but other than that I mostly used Python which is what it seemed like most people used. I feel like it could've been helpful to know and use Matlab for the class since the class was taught with it but it's definitely doable with any of the languages.

Each TA had several office hours they held each week and the ones with attendance got recorded. They were mostly Q&A sessions about topics in the slides or to go over exam problems. There were times where I felt like information from one TA was contradictory with another and had to wait a few days for that to get cleared up. Initially they did help people if they showed them their code but they eventually announced after the first 1-2 HWs that they wouldn't be helping with code but just understanding topics, after which the attendance for OHs went down by a lot. They only recorded sessions if people went so there weren't that many recordings by the latter half of the class but they were mostly answering questions on Ed. Also I think a previous review from 2022 mentioned a TA was pretty generous with their hints, I would say if you were to take it now that might not be the case and you should expect to search for resources and aid on your own.

There were occasionally typos in the slides and homeworks and there were 3 incidents I remember of assignments being released late ( 1 of which supposed to released on a Monday that was a Federal Holiday in the US and 2 being exams) and only the exam 2 had received an extension since there were issues with the files to be used as well.

Despite all the issues, the TAs were EXTREMELY lenient with their grading which I do appreciate. I think halfway through the class I was resigned to just accepting a C but somehow I barely ended up with an A. As long as you try your best to answer every question to the best of your ability, you shouldn't lose too many points.

In comparison to the other stats options, I don't think there were other classes I would've rather taken since REG has bad reviews and I prefer assignment based classes so I would probably take it again if had to pick again. I think the best way to take this though would probably be taking DO before this (which I don't plan on taking) or taking this in the summer so the optimization models get dropped. This class gets referenced as Machine Learning 2 a lot with CDA being called Machine Learning 1 and I'm not sure if that's the best way of seeing this, although the class was interesting it's more niche so I'm not sure if I'd use the content in this class as much as one might from CDA.

Lecturers were great with plenty of examples! Lots of intense maths, but the effort required to complete the homeworks/exams is lower than other courses like DC, Compilers, AOS, and DL.

This was my 8th course in OMSA (having previously taken all core courses except mgt 8803, plus SIM, DO, and CDA). GPA before the course was >3.5. I seem to have sneaked by with an A here, but it was not an easy achievement.

I chose to take this course in summer, as they drop the optimization modules to squeeze the course content into the shorter semester. I had previously taken DO and didn't really see the point of re-sitting that content. To their credit, the dropped optimization modules are available for study, and do admittedly add value on top of the content covered in DO.

I REALLY liked this course from a content perspective. The lectures, readings, and example code were all excellent resources. I truly regret that I didn't have time in my personal life to spend mastering this material, as it is mostly things that I had never come across before.

I have indicated 15 hours as the time I SHOULD have spent on the content to comfortably do well, but realistically, I averaged about 9 hours/week in the form of nothing during the week and then late, high stress nights and weekends trying to simultaneously learn the material and get everything done at the last minute. Not proud of my approach here, but it was a rough few months between life/work/personal health. I say all this to add emphasis to the point that if you actually average 15 hours per week of study consistently throughout the course, the lecture materials, homework and exams are all set up such that you can get an A in the course without killing yourself. No gotchas, no unfair / trick questions.

This is a great course that I am thankful I didn't let the higher OMSCentral difficulty score discourage me from taking.

Lots of great content… I really enjoy the way prof walk me through every each of them: first display equations, then explain each term in the equation sometimes with pseudo code, then one or two applications with plot visualization, at the end matlab example code. I also can find myself apply the techniques taught in class to my daily work in the near future (signal processing, analysis..). Only thing if there is any, I would like to get a bit more understanding of the formula in applications. In homework this is usually a given. I would like to know how these formulae are derived from real engineering cases.

Really interesting topics! The lectures covered the mathematical formulations and also provided programming examples of applications. Homeworks were fair, with a lot of TA help and class participation. Lecture slides are very well constructed.

I wouldn't say this course is really "math heavy" because all of the homeworks and exams are programming in Python or R. But the course very quickly dives into matrix operations.

I'm 8 courses in to the C track for OMSA, but I was still kind of worried about this class over the summer. However, everything felt very manageable. I think if I took it as #3 or #4 I might have felt overwhelmed at times, but a lot of the material in this course felt familiar to bits and pieces learned from all of my previous courses.

Great course that covers some more niche but also very interesting and insightful areas of machine learning and statistical analysis. Basic linear algebra (SVD, inverses of matrices) and calculus (derivatives of scalars and vectors) is required. I used python for the assignments but R and matlab seemed slightly preferred as the solutions are often in matlab. The topics covered range from functional data analysis to applications of regularization such as robust PCA and matrix completion. Every module was more interesting than the last.

Hold on to your hats, folks, because this course is a BEAST! It's the most mind-bendingly mathematical course at OMSA, bar none. Get ready to level up faster than you can say "linear algebra" and "machine learning." Professor Paynabar is a force to be reckoned with, delivering lectures of the highest caliber that cover everything from ordinary least squares to smoothing splines. But don't expect any hand-holding in this class - the TAs are happy to help, but you'll have to put your thinking cap on and work hard for every solution. And let's not forget the homework assignments - they'll leave you shaking in your boots no matter how much you prepare. But fear not, if you stick with it, you'll come out on top and learn more than you ever thought possible

Class is an enjoyable experience that transformed my fear of math into a newfound appreciation for it. Below are topics we learned for each module. Module1: High Dimension data introduction, Functional data (space vs time), Regression - Least square Estimates, Polynomial Regression, Splines, Order-M Splines, B-Splines, Smoothing Splines, Natural Cubic Splines, K-Nearest Neighbor, Kernel Smoother Regression, RBF Kernel, Functional PCA Module2: Image Analysis, Transformation, Convolution, Convolution with a Mask, Segmentation, K-means clustering, Edge detection using derivatives, Sobel Operator, Kirsch Operator, Prewitt Mask and Gaussian Mask. Module3: Tensor data analysis, Basic mathematical operation, outer product, inner product, Kronecker product, Khatri-Rao product, Hadamard product, defining tensor ranks, rank one tensor, Candecomp/parafac (CP) decomposition, Tucker decomposition Module4: Regularization, Ridge, Lasso, Non-negative, Adaptive Lasso, Group Lasso, Module5: Compressive sensing, using Regularization for CS, Optimization Algorithm for CS

This course is very heavy in math, very difficult and time consuming if you want to understand EVERYTHING. I personally, sort of scraped by because TAs grade very leniently, but if you want to truly absorb everything this class can take up to double the amount of time. I'm not exactly sure how useful everything taught in this class is compared to whats popular in the industry such as Deep Learning, AI, etc.

If I could rate the course content separately, I would give that a 5/5. However, the management of the course drags this down to a 2/5 overall for me.

DIFFICULTY: This is an extremely math heavy course, specifically with a focus on linear algebra. I took linear algebra in college about 8 years ago at the time of writing so I was very rusty but I managed to catch up by spending a lot of extra time Googling concepts. The Matrix Cookbook is very very helpful for this course. Wolfram Alpha can be helpful for the first few modules too but it quickly becomes useless once the math gets more advanced so you're better off just learning how to do it yourself from the start or else you risk losing your crutch halfway through the course.

CONTENT: Tons of interesting under-the-hood stuff is explored in this course, and very quickly. You get 2 weeks per homework and 1 week per take-home exam, which sounds like a lot but because of the difficulty is really not much sometimes. Working through the homework will give you a solid understanding of the concepts presented in lectures. Participating in the forum is a must and SOME of the hints that the TAs give there are very helpful. The exams are generally easier than the homeworks but you are not allowed to ask questions in the forum for them which can feel scary. Homework and exams are graded quite leniently, so as long as you can show that you have some grasp of the concepts. I have no doubt that the content is useful if you are going down the data scientist route, but it's probably not anywhere near as useful for business analysts, BI analysts, or data analysts. In my opinion, there are so many concepts introduced so quickly here that it will give you a good idea of what you need to learn next, but completing this course will not give you the confidence to apply the techniques in real life immediately.

LANGUAGES: Most of the examples in the lectures are given in Matlab with occasional R or Python. Example code is usually given in all 3 languages, just not shown in the lectures. You are allowed to submit your homework and exams in any of those 3, however I noticed that most students used Python, a small group used Matlab, and basically no one used R. This is important because it means you will get the most support from others using Python. However I used R and found that I didn't have as many issues with libraries and variable types as students using Python. Homework solutions are also given in all 3 languages, but for some reason exam solutions are only given in Matlab.

MANAGEMENT: This is the biggest letdown in the course for me. The professor is basically non-existent as I don't recall him interacting with the students in any way, ever, during the entire course except for showing up in the pre-recorded lectures. No office hours, no forum posts, no announcments, nothing. The TAs try their best...sometimes. If you post your answer (graph, coefficients, MSE, etc.) on the forum they will tell you directly whether it was correct or not. There was one TA who was quite generous with his hints and always happy to help, however other TAs tended to give vague answers to students' questions. There was another that gave straight up wrong answers several times without correcting himself after discovering the mistake which caused many students including myself to go on a days long wild goose chase. DO NOT assume that TA answers are always correct. This extended to exams too, where one of the crucial hints given in the exam was wrong, causing me to waste several days trying to figure out why my code didn't work.

Lastly, I found it incredibly frustrating that so much of the homework and exam grade is based on solving special cases of equations in otherwise straightforward questions, because it means you are essentially being graded on esoteric math knowledge not mentioned in the lectures. For homework this is OK because you can get help from classmates and TAs in the forum, but on exams if you're not already familiar with whatever random concept they decide to introduce, then you're gonna have a hard time. Again, hints provided in the homework/exam or by TAs ARE NOT NECESSARILY CORRECT, so if the hint doesn't make sense or you get stuck, try going back and assuming that you've been lied to at some point. If your Google skills are good enough, you may be able to find solutions from academic papers or old forum posts (literally, I owe at least 33% of my grade to 10+ year old quora posts).

Hands down the best course in OMSCS I have taken so far. This was my 7th course after ML, RL, DL. The course covers some fundamental theories of machine learning which I never came across elsewhere.

• The lectures were divided into 7 modules. The videos are short, yet covers the content in sufficient detail. The assigned readings complement the lectures well.
• The course deals with the domain of problems where data is of high dimension but number of samples available to train is very small (<100). Throughout the course, we did some cool problems like time series regression, image classification, image inpainting, image restoration etc. The algorithms had well rooted theory, so no black box, and most could be trained under 30 seconds (some less than 1 second).
• The assignments were well thought out and cement the concepts taught in the relevant module. While some questions directly follow from the lectures, some required deep dive into the reading materials. Most of the problems required to first come up with the derivation/proof for the problem statement and then code the algorithm out.
• TA's were super helpful and active on Piazza. The OH were held multiple times a week to discuss the assignments and the TA's even helped students to debug their code.
• The assignments can be done in MATLAB, R or Python. We were allowed to submit our Jupyter notebooks/R markdown/MATLAB live script directly as report. I did all the assignments in Python except the one on tensors which I did in Matlab.

This course is definitely hard but doable and highly rewarding. I was skeptic coming into the course because of its difficulty and elaborate prerequisites. A background in linear algebra and multivariate calculus is required but not at expert level. The grading is relatively on the lenient side and you can bag full credits even if the answers do not match with the TA's solution as long as the approach is correct and reasonable. I found The Matrix Cookbook as an invaluable resource. This course has absolutely changed my outlook towards ML in general. Not to mention I am no longer scared of the math of machine learning.

One of the best course I have. Professor Kamran 's office hour is a must join if you have time. Professor is patient enough and will explain the lecture and homework during his office hour. He is also willing to go through some deeper topic if you want to.

I did everything in Python. TA has provided example code in R, Python and Matlab in this semester. There are some error in example, but it is not terribly hard to debug those and get the homework done.

The optimization section covers several applications -- gradient descent on parameter , image separation ..etc. it covers both first order and second order method. The method can be used on constraint optimization as well

The tensor section is definitely new to me.

Overall the course did broaden my knowledge in analysing high dimensional data

A lot of math inside, be prepared.

Strongly liked the course.

However, the removal of the 2 modules in the Summer makes this course way less challenging than it should be in Spring/Fall. I saw the 2 skipped modules (Modules 4 & 5 in Optimization), they seem to be way more difficult than, say, Image Processing in Module 2.

Suggestion - In summer to remove Module 2 (since it's also Matlab-driven) and replace it with Modules 4 or 5 to make the summer equivalent as challenging as the regular semester.

This helps to protect the integrity that this course is reserved for the best of OMSA.

Enjoyed the course because it dug into some tools to use for HD data if you aren't at the Big Data amount of data. We covered regularization, image process (e.g., convolution), tensors, splines, and functional data analysis. Summer semester kept the optimization module as optional and provide lectures, hw, and hw solutions for that module. The course dove into the math side of the tools, but was really good about providing example code for homework problems and grading seemed to be somewhat lenient. I was usually able to complete HW assignments just by working on weekends. Also, the exams were take home and very similar to HW assignments.

Cleared due to OMSCentral Owner being greedy.

HDDA is the most mathematical course I have taken in OMSA. Make sure you are very comfortable with multivariate calculus and matrix algebra before diving into this course. There are lots of gradients, tensors, and other fun mathematical topics. The algebra goes well beyond the typical one semester undergraduate linear algebra course taken at American universities. The HW and exams include rigorous math proofs, and even the programming often requires you to mathematically derive the algorithm that you ultimately code. I strongly urge you to do the assignments in Matlab, as it is much better than Python for implementing the matrix algebra that is used in this course. I did a few assorted HW and exam questions in R, but the bulk of my coding was done in Matlab. If you don't know Matlab, don't be scared. I hadn't used it before, either, and I picked it up quickly. The optimization modules were by far the toughest part of the course, and they are cut from the summer version of the course, which could be an advantage if you don't like hard math or a disadvantage if you are hoping to gain optimization experience. I would say this was the hardest course I have taken so far, but I liked it a lot.

Expect to spend many hours on the HW and exams. The exams are take-home and un-proctored; we could use any resources except other people while taking the exams, which made them less stressful than they could have been. HW typically had ~2 weeks to complete, while exams had ~1 week. The most stressful period of the course was when HW4, Exam 1, and HW5 all followed each other within a short interval of time.

The TAs were excellent, and the lecture videos were some of the best I have seen thus far for OMSA. This is my 9th course; I am on the Analytical Tools track.

This was a great course - one of the best experiences I've had in OMSA (6 courses in, but haven't taken an MGT course yet). I'd highly recommend this course to anyone considering it. For benchmarking, I also really enjoyed ISYE 6740, enjoyed ISYE 6501 and ISYE 6444 (Sim), thought CSE 6040 was decent, and found ISYE 6414 (Reg) useful but disliked the class.

I found the material on splines, tensors, and compressive sensing all to be very interesting. There was some overlap on the optimization and image content between this class and CDA.

The class is tough. The material can take time to get your head around. But the example code is pretty thorough and the TAs provide a lot of help and guidance through the HWs. The exams can be stressful with the safety blanket of Piazza/TAs gone.

I struggled at times with the derivations and wish they provided better resources to help students whose math skills might be rusty or not quite where they expect them to be.

Concerning time commitment, I'd estimate that, for each module (7), I spent 2-3 hours watching lectures and up to 20 hours on HW. I spent 25-30 hours on the first exam and 15-20 on the second. I didn't spend much time reading the reference papers, usually only if I were struggling with a topic. The high side of all those estimates works out to ~14 hrs/week. If you put in the time (and it might be a lot of time), I feel like it will translate to learning the material and getting a decent grade.

Finally, some parting advice: get comfortable with matlab from the start. I think the class was originally built on matlab, with R added as an option later. But things in R didn't always behave the way they do in matlab and the final module doesn't have example code in R.

This resulted in 10 wasted hours spent troubleshooting on the first exam when I could've just switched to matlab (I think it also cost me some points). I eventually switched to matlab for the last HW anyway, which made that HW much easier.

• tl;dr 1 : This class is not for the faint-hearted. If you don't have strong Math prereqs obtained in CDA and DO, you're going to have a rough time, even for summer.

• tl;dr 2 : Don't be Python-stubborn. Learn Matlab.

They say this is the mother of holy grail of OMSA for a reason. This is not a normal brick and mortar class. No textbooks will ever sufficiently prepare you for this class. Forget about watching these videos in advance. You won't understand much until you put your hands dirty on the homeworks.

You need a combination of Linear Algebra (a dose of ISYE 6669 preps you well), Calculus III and do well in it, Machine Learning theories (some doses of ISYE 6740 or 7406), and a hell lot of grit to get this through. So make sure you plan your courses well prior to taking this class.

The Mathematics you understand here puts you one step closer to doctorate standard. You will be splashed with cutting-edge knowledge that you will be given tons of research papers to read and implement - some of them written in English but as a second language so you will have to understand their linguistic context as well.

Codes are given sometimes in R, mostly in Matlab, but for a reason. The TAs and the Prof believe that such particular theory works the best on that program. The time spent may vary, owing to debugging, but the answer codes are usually the most elegant.

Feel free to port to Python if you really choose to be stubborn. It really pays off to be programmatically multilingual - not to say that Python is horrible at handling high-dimensional data types. After all in the real world, you work with programs that give are the most efficient for that algorithm.

This is easily the most intense but fulfilling class in the entire OMS program. The good thing also is that you probably won't get to hear annoying messages in Piazza and Slack because the course is so intense, those whiners weed themselves out in the first week.

Many of the good aspects of this course have already been covered accurately in below reviews, so I am going to focus on what hasn't been mentioned too much -> the negatives of this class.

I felt that the pacing was too quick in the middle portion of the course. Even the top achievers of the class were rushing to finish homework 4 and 5. Normally, we get 2 weeks exactly to do the homework, but for whatever reason, homework 4 gets 1.5 weeks which is directly followed by 1 week for the exam and then homework 5 gets 1.5 weeks. I felt my survival instincts kick in and I focused less on learning the actual content and more on just trying to push through to get the answers correct on the homework. This had the domino effect of adversely affecting my speed to complete later homework assignments, because I did not take the time to really learn the concepts well.

Next, I should mention the flaw in current office hours setup. We have 6 office hours scheduled throughout the week. I personally found myself as an office hours frequenter because Piazza quickly became a web that had to be untangled and not all points are addressed in Piazza (there are a LOT of smart people taking this class, so pretty much no questions on pre-requisite material used are asked). I myself spent the majority of my weekend doing the homework, but that only left me two office hours to ask questions, and who wants to wake up early on Saturdays? I'm on my own after that. I wish we had moved some of the weekday office hours to double up on weekend (i.e. 2 OH on weekdays, 4 OH on weekends), so that better dialogue could occur on office hours.

Also, this class is kind of weird in that it jumps around between Matlab and R. I decided to learn Matlab during this course and Matlab was the best option for the majority of homework assignments. I think that this class would benefit from having answer keys and example code provided in all three languages that it supposedly supports -- Matlab, R, and Python. This would truly give the class the flexibility that a Masters class at Georgia Tech should have.

And warning to all who take this class, TAs often do not read comments on private threads unless they are recent. Just create a new post and it is more likely to be read.

Finally, I'll mention why after all that I decided to put "Liked". This class challenged me greatly. It taught me so many cool concepts. It gave me an opportunity to message with brilliant minds. Despite all the struggles mentioned above, it was a fun class. Be prepared to learn some advanced math concepts, and buckle up for a bumpy ride.

This course was recently opened up to OMSCS(previously only as an OMSA course) and I am glad I signed up. The syllabus and contents peaked my interest enough to sign up. This is my 7th in the program and I'd rate this as one of the best run classes I've taken. There are two take home exams for the mid term and finals, which is perfect you are not a fan of closed book exams.

The material is well curated as in you only have about ~1 hour worth of weekly well-explained, videos that breaks down the content. The hand out slides of the same are also provided along with a set a example codes. These are very helpful for the bi-weekly 7 homeworks. You can do well on the exams if you were able to figure out the homework problems.

The homework consist of 4 questions which are typically 3 coding problems and 1 proof. The first three homework topics covered functional data analysis using Splines, 2D Images(nothing new if you have taken CV), Tensor data analysis and decomposition methods. From HW 4 onward, the topics got progressively more complex- Optimization(Gradient descent, decomposable functions like lagragian, ADMM) and Regularization(lasso, ridge, elastic net,etc..) methods, Compressive sensing, PCA. For the most part you are expected to able to derive the math to set up your solvers. So you are expected to be comfortable with partial derivatives and linear algebra.

The TAs were very helpful and approachable throughout the course on Piazza and the Office hours. There were ~ 4 office hrs every week. They really took the effort to make sure to no one was left behind. The prof. seemed approachable and would occasionally post on Piazza. I'd recommend this course as a follow up for anyone in the ML specialization.

One caveat, there is no fixed programming language stipulated however 90% of the sample codes are provided in MATLAB. You might be better off if you are comfortable with Matlab. You are encouraged to do this in the language of your preference and I noticed several in the class were doing this in Python and R.

Before getting into the review, I'd like to provide my previous coursework to give context: CSE 6040, ISYE 6501, MGT6203, CSE6240, ISYE 6669, CSE/ISYE 6740 (A's in all classes). I also took this class in combination with simulation, as I waived out of MGT8803.

I loved this class. The topics were definitely more advanced than CDA and it felt like a machine learning II class for the most part. There are proofs in every homework assignment (just one out of four questions) and exams, but the remaining 3 questions could easily also require a derivation before implementing into code. The TAs were excellent, providing very quick and thorough responses. Also, it felt the students in the class were extremely smart, where most were part of the OMSCS program with courses like AI, ML, ML4T already under their belt. There were not any questions that felt out of place on piazza. If I didn't take my heavy computational workload before this class, it would have been a struggle to excel. It starts out with material of which is considered "appendix" material in ESL, B-splines. These are referred to throughout the rest of the course. Then basic computer vision, tensor decompositions and math, optimization (first section felt like review for me as I took ISYE 6669), but the second section was amazing. The second section covered state of the art techniques used today for optimizing ML algorithms like coordinate descent, ADMM, etc. It then closed with sparse datasets of high dimensions, highly reliant upon the Lasso framework and applying it elsewhere, compressed sensing for example.

I thought the teaching was amazing in the videos as well. The math is absolutely not hidden away from you, and the teacher does a great job walking through exactly what is being shown. He will never just read off the slides without walking you through the material. The amount of videos every couple weeks is lighter than most courses which is nice - most of the work is in doing the homework, doing the reading (either what he gives, or from Stanford/CMU/Princeton. Just like ISYE 6740, I relied a ton upon these university's materials). The references were almost always papers - that's when you know the class is legit.

Sometimes, it feels great getting the problems correct. They seem very scary at first, especially getting into things like matrix completion with a higher-dimension and sparse dataset, but you are armed with the tools to solve the problems. I always referred to the class as the "beast class", due to the respect I had for the material being covered; a first in the degree for me. I loved the applicability to real life - as a data scientist in real life, you deal with sparsity ALL THE TIME. Concepts like tensor math + decomposition will serve me well in the deep learning class.

If you are not very adept in programming in any language of your choice (students can use MATLAB, R, Python), don't have a solid background in linear algebra/multi calc/stats, and haven't taken a ML class beforehand, this class might be a kick in the mouth. The optimization class definitely helped, but I felt the material was taught well enough where you could fill in the gaps if need be here. ML classes cover optimization already. Concepts like SVD/PCA/Kmeans are assumed that you understand fully in depth - you are required to build both a tensor decomposition algorithm from scratch, basically a higher order SVD (HOSVD for short) with a ALS set-up, and robust PCA to handle sparsity. Hence why I say it is more of an ML II Class. From reading previous reviewers, and seeing previously given homework assignments, it feels like a conscious effort was made to make the assignments more rewarding and less reliant upon the example code. It felt like the example code was designed to be what it was - just an example to get your feet wet. I don't know for certain without comparing fully, however. To provide substantiation that the material may be different now, the average on assignments usually hovers around an 80. Highest i've seen is an 89 for a homework, lowest a 70. Definitely not a walk in the park for a generally smarter student population!

I would highly, highly recommend this for any student in the computational track in OMSA or the ML track in OMSCS if you are more seasoned in the degree. It is a class size of around 75-100 students with amazing material in terms of teaching and assignments, especially if you like math.

I do not recommend taking it in the summer, as the optimization modules feel like a must. I just finished the final exam - both the midterm and final felt very fair and like a homework. I am not giving a very hard rating because some problems were more of the CDA style of running a model in a package of your choice, but those were few and far between. It would take a ton of time to even get to that point, and this was very occasional.

The example code given is amazing - I'm sure I'll rely upon it throughout my career.

Decent course, glad I took it in the summer. Some interesting topics not covered in other classes, except the image processing stuff, which is covered in CV. While they skip the optimization modules and leave them as optional, the workload of this class is light enough that you can easily watch them/do the hw on your own during the summer. Workload is comparable to something like ML4T, although the content is much more advanced. Don't worry about the "proofs" you hear about in the homeworks. They are more like derivations of formulas (i.e. very informal) and typically just require some algebra.

HDDA was my 5th OMSCS course after AI, RL, CV and ML. I liked it because it covered quite a number of topics that had not been taught in the courses I'd already taken, such as splines, elastic net, optimization techniques, tensor math and tensor decompositions. The image processing module didn't really teach anything new that wasn't covered in CV, but it was also the shortest module.

Although there are no official pre-requisites for this course, I think a basic grasp of linear algebra should be considered necessary. Moreover, although it's not too hard to get an A, the stronger one's math background, the better one can really master the material. I benefited a great deal this semester from some of the other students who knew more math than I did. This was in part due to the 5% extra credit that's given for participation in Piazza, though it also had the unfortunate side effect of incentivizing students to give big hints for the homework problems, since only answers were eligible for the extra credit.

There were some mistakes in the lecture slides that apparently previous cohorts had already pointed out, but were not yet fixed due to poor version control.

One downside of taking this course in the summer is that the 2 optimization modules were made optional. I would have preferred if they had cut out the image processing module instead and reduced the amount of time given to the other modules in order to keep the 2 optimization modules in.

(omscs studient) Amazing content, best lectures very concise and interesting, exams and homeworks are very rewarding.

This was my 2nd class in OMSCS after RL, and my 1st ISYE class. Taking this in the summer, one of the modules (optimization) was skipped entirely, so if this matters to you, make sure to take it in the fall/spring. The class covers topics very relevant to ML - spline interpolation/regression techniques w.r.t. high dimensional data, image processing/computer vision, tensors, regularization.

Some reviewers say the content that is too old, but I've heard that much too many times and think it is unfounded; it is important to understand the building blocks that came before, since they often cover the foundation for newer techniques.

The lecture content for this class is great - the delivery of the lectures by the professor is alright, didn't have any problems, but never thought he explained concepts perfectly or that greatly. However, the mathematical depth in each lecture is very much appreciated, and what I enjoyed the most; however, the depth is only enough for more intuition on how the models work, and not as in depth as I would have liked (some lectures are more in depth than others).

The downside to this class is that it tries to compress too many topics into one class, so they never really get a chance to go too in depth on one topic.

This downside is seen vastly in the assignments/exams. Despite my enjoying of the content of the class (but wanting more depth), the quality of the assignments is why I give this class not a superb rating. The class provides you with example code that is run in the lectures, and 99% of the homework problems/exam problems that aren't proofs are literally just running the example code again with different inputs and bit of tweaks - this is not stimulative to learning to really understand the lecture material, as the content is black-boxed in the example code.

Additionally, the TAs literally hold your hand on piazza for every homework assignment, and assignments were mostly a problem of figuring out what the TA wanted you to do, instead of you applying knowledge that you've learned from the lectures. (Exams are the same format as homeworks, take-home, but that nobody can help each other)

On the flip side, I enjoyed all of the math proofs very much, but the lack of breadth/depth of the math in the course made it not proof-heavy and more application heavy, so a few proofs here and there didn't really teach me enough to understand how it relates back to the models as a whole.

Its true that you can just read more of the math content in depth if you want to really learn more. Its also true that you can just write your own code to really learn how to apply these techniques instead of using their example code. This class does give a good opportunity to learn these things in depth on your own, if you desire. However, the way the class it set up right now, you can also pass the class with an A (I believe I am getting an A, and the median on all assignments is an A) without really retaining much of the material. Depending on your math background and how much of it you really want to understand and how much you just want to get done, this class can be very easy and unsatisfying, or very difficult and satisfying.

Course emphasizes learning some niche but less useful techniques. Course material looks like it has not been updated in over a decade. A bit of a joke given how modern ML approaches dimensionality reduction. Focus on using R and Matlab is annoying. Ultimately there are more useful ML courses out there.

The one good thing about the course is that it does emphasize the math. A lot of other courses skim over the math but this course really goes in deep into the math portion.

8803 is a tough course, but not hard to achieve an A. You have to code from scratch and complete some math proofs, but it is not impossible. We only had two TA’s but they were both helpful. The lectures were good and free of mistakes.

We had 7 homework assignments that were due every two weeks (but it only takes about a week to complete them). The exams are not bad; the only issue is sometimes it can be confusing about what the question is asking. You could pair this with an easier class like ISYE 6420 or MGT 6203. I would not pair it with ISYE 6740 or ISYE 6669.

Overall I enjoyed the class and would recommend it.

TLDR; If you can handle the math you should take this class.

The good. This class is very challenging and exposes you to some cutting edge stuff that you won’t see anywhere else in OMSA. Things like Tensor Decomposition, matrix factorization, robust PCA with images to separate the background and foreground, compressive sensing and image processing. It is super cool and there were times when I was done with my homework and the result just felt like magic. This class is insanely difficult but it pushed me to learn so much and for that I am grateful. Now that I am at the end awaiting my final grade (good shot for an A) I am glad I did it but that is not how it felt during the semester.

Now the bad. This class has no professor. Technically it does, you see him in the videos and he will occasionally respond administrative questions related to the course (what time are office hours, when will the exam be released,etc.), but I cannot recall even a single instance during the entire semester where the professor of this class publicly answered a question related to the content of the course on Piazza. At one point the head TA admitted that he hadn’t spoken to the professor for weeks. It is honestly bizarre to me to read reviews from previous semesters talking about how responsive the professor is.

So between an MIA professor and the very quiet piazza forums, you are left to get any help from the two TAs to the course. The head TA is a PhD student who is generally very patient and helpful and your best bet for help with the homework. The second TA is a former student of the class and he uh, left a lot to be desired. He scheduled his office hours at 9pm to 10pm eastern time ON A FRIDAY. No only that, the couple of times I did attend it was a waste of time. He was generally unprepared and since he took the class over the summer didn’t do the optimization modules so could not answer my questions on those assignments. The TAs in general are solely focused on answering questions related to the homework. If you have questions relating to the course that don’t specifically pertain to the homework they can’t help you. The responsiveness of the TAs is also inconsistent. As I look right now (the exam is due in two days) there are questions open relating to unclear instructions on the exam and no one, not the professor or the TAs have answered them in over two days.

If you want to succeed in this class, you need to be prepared to do it without much assistance. If you are a fool (like me) who spent his undergraduate education aggressively avoiding any kind of math, this class will most likely be too much for you, unless you are prepared to make this class almost a full time job (I had to). If you do, you’ll come out the other end better for it, but you will suffer for it.

It was a tough course but also satisfying! The TA's were very helpful and patient in answering questions. I learned a lot from this course! I would highly recommend it but do take note that a lot of time would be spent on reading and answering homework questions.

Some prior mathematics knowledge on linear algebra and calculus would be helpful!

• This is a very hard course, math proof heavy.
• Homeworks are harder than exams, if you did well in HW, you can apply some concepts in HW to exams problems, so do take advantage of office hours/piazza to learn more/deeper about HW, it will be very helpful in exams.
• I took all ISYE courses except det. optimizatio and took a lot of AI/ML related courses when I was in OMSCS program before I took this course, with those courses under my belt, this course still felt hard.
• Overall, challenging but highly rewarding course, got to read lots of papers and learned lots of new concepts I never learned before.

Holy moly, what a class. I really enjoyed this class. Learned a lot and the TAs were top-notch. However, I definitely recommend you take it later in the program, especially if you are lacking in math background. This class involves some pretty complex linear algebra applications.

There was a homework assignment due roughly every 2 weeks. These tend to be very challenging and include written math proofs as well as coding exercises. Though you can do homework in any language, the instructor provides sample code in Matlab or R, so it's generally easier to just rework the given example.

The exams are basically just like homework assignments - you get a week to work on them and you can use the internet or other resources, just not your classmates or Piazza.

ISYE-8803 HDDA is a beast

Primary contributors to beast status:

1. Volume of material - So many concepts covered
2. Depth of homework - HW every two weeks, on average, required many hours
3. Mathematical foundation - There were proofs

This was a good capstone class that solidified many concepts from the OMSA curriculum.

Homeworks were 3 to 4 problems that each went into depth. Solutions were submitted as PDF write-ups, with source code attached. There was no weight given to language or implementation details, but all examples were provided in either R or MATLAB. Examples were very helpful, sometimes extending beyond the examples provided in the lectures. I know of some who completed all homework in Python as an "additional exercise", but due to time constraints, I stayed with R/MATLAB, typically modifying/extending example code.

Two exams, mid-term and final, were basically "take-home exams" in the same style as the homeworks, but with reduced reliance on peers and instructors, but open-everything, minus collaboration. Whereas the homeworks were 2 weeks, the exams had closer to a single week.

TAs were very helpful, with many office hours recorded and very quick Piazza replies.

The instructor was very responsive, both via Piazza and email, often "beating the TA to the answer."

There were only 38 of us in the class this time, which made collaboration more difficult since we weren't quite at critical mass to have someone else in the class always "in the same situation at the same time."

Overall, I highly recommend, but take it solo or paired with a lighter course (than CSE-6250 BD4H, that I inflicted upon myself).

The math in this course was SIGNIFICANTLY harder the other classes I had taken in the program (foundational + MGMT 6203). It took me a while to get past the intimidation factor of the notation alone. After a while and many hours of watching lectures, reading and re-reading supplemental materials I was able to finish the homeworks/take home tests.

PROS:

• Very math heavy and you will get comfortable with more complicated notation and derivations that many of the other courses I have taken lacked. First course that felt "graduate level" to me.
• Interesting material covered such as image processing.
• Can make sure you get an A by doing all homeworks and tests, but they take a lot of time

CONS:

• Concepts such as Tensors are more "niche", can be very difficult to grasp with little information outside academic papers, and may require significant amount of effort to begin to understand.
• Summer was very condensed and included optimization as optional material which I think is one of the more relevant topics for analytics. I would recommend taking this during a full semester so you cover this.

Homeworks have a math derivation portion and a coding portion. Took me a very long while to complete them. Most of my time was spent attempting to understand the topics and figuring out where to begin. I would recommend starting early.

Midterm and Final are take home and are in the same format as the homeworks and you have 1 week to complete.

Not a class for the faint of heart. Course was by far the most difficult in terms of mathematical rigor - definitely learn Greek letters and learn how to translate formulas into English. The next most important thing is linear algebra comfort and ability to track indices - as the course title suggests, you're dealing with an insane number of columns and rows of data and using linear algebra is much more efficient than looping. This is the first time since beginning (and appropriately, the end) the program where the course felt like the difficulty hit the graduate level. Nevertheless, as larger data sets and more predicting variables are made available, the topic will become increasingly important in Analytics as a whole.

In the Summer semester, the optimization modules (2 lesson sets/homeworks) were cut out entirely but made optional. Regular semesters will see these in the curriculum. Overall, there are 5 or 7 modules and homeworks to complete (x2 weeks or so each) over the course of the semester. Grading is fairly lenient and mostly effort-based, so if you put in good effort and come to an answer that makes sense you should get most of the credit. However, note that learning is facilitated by making an active effort to understand the code provided and reconcile it against the lectures. Otherwise, copying/pasting code from the lessons to apply to homework will work to your detriment during an exam if a curveball is thrown.

Breakdown for the Summer semester was 30% homework, 35% midterm 1, and 35% final exam. Exams are take-home/unproctored and basically homework problems with mild steroids and variation, but are fair and will not be on topics never before seen. Grading is similarly lenient. Additional 5% is given for active piazza participation.

Only piece of feedback for improvement on this course would be actually commenting the instructor-provided code. However, given code-reading is a practical skill, maybe this is by design.

This is an amazing course, and I would strongly recommend any OMSA student to take it. The course has total five modules. They are as follows:

1. Functional Data Analysis: Concepts of Splines and BSplines are introduced in this. Spend as much time as possible on this, and grasp this completely. Splines will haunt you the whole course :)
2. Image analytics: Concepts related to Image processing.
3. Tensor Data Analysis: Very interesting concepts. Linear Algebra is a strong pre-requisite for this module
4. Regularization: Linear Algebra and Optimization techinques are assumed to be pre-requisites.
5. Applications of Regularization: Linear Algebra and Optimization techinques are assumed to be pre-requisites.

It is assumed that you have prior knowledge of Linear Algebra; You can complete the course without Linear Algebra knowledge by spending extra hours on the topics as they come, but I feel it won't be very effective in terms of long-term retention of the material. Also, the last two modules rely heavily on optimization techniques. So if you plan to do Deterministic Optimization course, I suggest you take that first.

You need to code in R and Matlab for the course.

This was my sixth course in the OMSA program, after completing 2 courses per semester in Fall 2019 and Spring 2018. Boy, was this intense.

The good:

• You will learn a lot. A lot of the assignments focused on a research kind of implementation, which I believe to benefit my learning quite a bit.
• Relatively easy to score an A in this course. Just keep on your toes and following Piazza discussions.

The bad:

• The reason for liking this course, is also the reason for disliking it. Because it was so heavy on research papers (which are deep and take a long time to read), the scope that was covered in class was limited. This placed the onus of covering materials on the students, which was a fairly broad task.
• Instructor gave a 5% bonus mark for Piazza participation, which I felt was unfair, as it encouraged adverse participation (ie people who posted for the sake of posting).
• TAs refused to consider different perspectives, even though there are different ways to interpret and solve a problem.

Very tough course, but also extremely interesting. As others have mentioned, this is likely the most Mathematically challenging course being offered by OMSA up to this point. The grading is very lenient, but you will still need to put in a lot of effort if you want a decent understanding of the material. Unfortunately, the homeworks can be overly vague or the notation is not explained well enough, and you will be forced to depend on fellow students asking good questions during office hours and you'll have to monitor Piazza constantly. Overall, the TAs for this course were the best I've ever had and we're extremely responsive on Piazza, and great at clarifying the homeworks

Honestly, my least favorite class in the program so far which is saying a lot because I really hated Business for Data Analytics. First of all, the example code was HORRIBLE. Whoever wrote the example code needs to learn how to code. First of all, nothing is commented and there is variable replacement throughout. Secondly, the example code goes back and for the between matlab and R so you never truly understand the material because you are constantly jumping back and forth. I had experience in both programming languages prior to this class, but when you are learning new material jumping back and forth doesn't make it easier to learn....The homework assignments are tough. All homework are 3 problems and most of the homework assignments consist of at least one proof problem. Make sure you ask a LOT OF clarifying questions on the test, because sometimes the problems are not worded correctly and wrongly interpreting will cost you a lot of points. Out of all the time and energy I put into this class I really don't think I really have a strong grasp on the concepts. I'm very disappointed in this class after reading all the amazing reviews because you can do the homework and tests by mostly using example code or googling for concept answers for the most part.

This is a very challenging course for me. It assumes a lot of prior knowledge on Linear Algebra and Calculus. The generous use of advanced mathematical notations to explain concepts behind feature selection, regularization etc makes it even harder for students (like me) with average mathematics background. I spent a lot of time reading up a lot of extra materials (mostly in mathematics) in addition to the recommended course readings and lecture slides.

Homework questions are a mixture of questions on mathematical proofs and programming. The professor allows students to use any programming language of their preference but example codes are mostly in Matlab and occasionally R. Using Matlab in the homework will certainly save a lot of time as example codes are provided and some Matlab functions does not have an exact equivalent in R or Python. It is easy to pick up Matlab if you are already familiar with R.

Professor and TAs are very responsive in Piazza and office hours. I would say both platforms are mandatory tools for homework and exam. The great student community also help a lot.

Overall, I would say this is a great course though very challenging. I rate this more challenging than CSE6242 DVA. I will not recommend taking this course in Summer semester as they adjusted to the shorter timeline by dropping 2 essential topics. Professor tries to include this 2 optional topics in the last homework which attempts to cover too much ground.

This course is probably the most challenging course in the program in terms of Math content. I strongly recommend to brush up Linear Algebra and Multivariate Calculus before taking it.

What I liked about this course is that the content goes very in depth in the content. In fact after taking this curse I realised other courses like Regression and Optimization could go deeper in the Math. The homeworks are very challenging and force you to get your hands dirty with equations and proofs without foegetting about the practical applications which as you might guess turn the homeworks into long assignments.

What I didn't like is that I really couldn't catch up with all the content and I think this is due to my own time limitations due to a full time job but also due to sometimes a teaching approach that assumed things as known where more intuition would have been appreciated. One more thing that could be improved IMO is that the lectures/HW are far from self contained and Office Hours become critical and much time need to be spent to figure out some concepts or requirements.

This course is definitely my favourite course so far in the program. Previously taken 3 core electives, 6203, audited time series and data base systems design.

The course is fairly math heavy, coding component is quite easy. Taking optimization first would make this course much easier. The knowledge imparted here can be quite practical, and definitely very interesting. The professor is excellent. Clear explanations and the instructors are very responsive.

I would highly recommend considering this course for your OMSA stats elective or as a track elective for the analytics tools track.

I loved this class, but it was very challenging. It consisted of 7 modules/assignments and two exams. Each module was two weeks long and consisted of 1-2 hours of video, some sample questions, optional readings, and an assignment. Each assignment consisted of a mix of mathematical derivations and coding applications. Exams were like the homework assignments (unproctored, similar questions), but you only have one week to complete (and you should start early, it takes a lot of time!). There was no curving of the final grades, but the instructors graded leniently - if you made a reasonable attempt but didn't necessarily get the exact right answer, they still gave you full (or mostly full) credit for the problem.

I highly recommend you take this class, but make sure you budget time appropriately. You may need to beef up your linear algebra/calculus skills before taking it; I found the math to be the most challenging part. I also highly recommend using MATLAB/R in this class to save time - the professor provides sample code and it saves a lot of time if you modify his code. (He also switches to R for one application, LASSO/ridge regression, and it's because R handles it much better than MATLAB... use R for those questions.) If you're able to finish the assignment, you'll likely do well on it; mean homework scores were high.

This could work as a summer class, but I don't recommend taking it with a second class if you're working full time.