Glosses over material (much like prior courses in this specialization), the professor is audibly nervous during recorded lectures, and many assignments require information and functions not covered in the lectures. Additionally, out of date Python modules are used in the notebooks, so you're learning often deprecated usage patterns, not to mention the constant struggle that is the auto-grader. You can teach yourself with free resources and save yourself the money and unhelpful bouts of rage against the auto-grader.

Not very good compared to the first two courses :( :( :( ... I took a Machine Learning Class from Stanford which was incredibly well put together and presented (though to be fair, it was 12 weeks), but it was in MatLab and I wanted to take a course in Python just to have a different perspective and solidify my understanding. Unfortunately, I find this course to be confusing more than anything. If I hadn't taken the Stanford course before, I'd be completely lost. It's very dry, dense, and hand-wavy and doesn't go into a whole lot of details with anything leaving you wondering what's happening and why and how... I don't approve of jumping straight to using the built-in functions if you don't understand the processes behind them (which I personally don't have a solid grasp on them still) ... I think they are just trying to fit too much information into four weeks and it's really lacking. Maybe if you're already familiar with linear regression, it's not as hard to follow. Either way, I'd recommend either taking the Stanford class first, or learning about this stuff elsewhere before starting this course.

There is a huge difference between teaching / tutoring and just reading some pre-written scripts. Even on an online course. Andrew Ng's Machine Learning course is a great example of teaching and this was one of the worst courses I have ever taken in coursera / udacity.

A lot of stuff, compressed in a short time. It's more about memorizing a lot of concepts rather than understanding them. I strongly recommend to take the course of professor Andrew Ng before this one.

I want to give this course a higher score because I do think I learned A TON. However, I learned a ton because the course had some flaws in instructions and assignments that required some frustrating moments and a lot of outside work to correct. If you take this course, DISCUSSION FORUMS are a must because of all the errors and bugs in assignments. The explanations are a little 'too rosy' in the videos in my opinion (they show best case scenarios) so there's a disconnect in what i actually had to do to pass the assignments which tended to have lots of room for improvement. That said, if you are willing to go out on your own and figure it out (mentors are so-so in actually helping), then this course is a great ML workout!

This course is ideally designed for understanding, which tools you can use to do machine learning tasks in python. However, for deep understanding ML algorithms you should take more math based courses

I am unenrolling because of the following reasons:

1) instructor lead training is very very boring - the gentleman keeps talking in same pitch and there is no lucid explanation behind the math that is constantly thrown at you

2) the course does not bother to put in any real world scenarios to correlate the content with

Overall really poor experience

This is a great course for those with limited experience of machine learning, wishing to quickly grasp how to apply machine learning methods and get their hands dirty. In my opinion, this is the best course in the specialization so far and as in previous courses you are expected to dig into further theoretical/usage details yourself from online documentation (hence the name applied). Concise lectures and interesting reading materials, as well as hands-on assignments. My recommendation is to either start with this course or take it together with more theoretical courses (such as "Machine Learning" from Stanford or "Machine Learning Fundamentals" from UCSD) to get the full flavour of what machine learning has to offer.

TLDR; Boring and unstructured courses that do not offer insight. You learn by doing the assignments.

The video lectures are boring and unstructured. You can tell the lecturer really hates what he is doing often sulking and showing zero enthusiasm. Also, he makes you question if he really knows what he is talking about. I am sure he does but his attitude and sloppy mistakes give you doubt. The format of the video lectures is that the lecturer reads a script in front of the camera and the algorithm he talks about is shown in cutscenes. This is a terrible idea. Also, the courses are not well prepared, lacking continuity. On top of that lecturer often makes mistakes and these mistakes are "corrected" by showing you a cutscene that writes the professor wanted to say X instead of Y. This is really sloppy. This is not an open course where you put your recorded lectures to youtube for free. You are delivering these lectures to paying customers. Seriously many free lectures on youtube are better built compared to these lectures. I have learned a lot of things in these lectures by doing the assignments and trying to learn by using google and not via lectures. One positive thing about this course is that there are some good links to papers, websites etc... But you need a lot of time to go through them.

This class is useless. There are countless errors that go unfixed in the lectures and autograder. Apparently you can't update anything on here because they haven't. Also while machine learning is an interesting and emerging topic, the lecturer makes it sound like old news and shows no enthusiasm when presenting this course. There are better places to learn about machine learning, for example Google has a FREE online course that I used to help me while doing this course.

Coursera/Michigan needs to review this course because it is not up to any standards.

The course was really interesting to go through. All the related assignments whether be Quizzes or the Hands-On really test the knowledge. Kudos to the mentor for teaching us in in such a lucid way.

The content of teaching is a way too less than the assignment's level. I had to make efforts on my own .

Kindly increase the content of teaching.

Very well structured course, and very interesting too! Has made me want to pursue a career in machine learning. I originally just wanted to learn to program, without true goal, now I have one thanks!!

My biggest critique of this class is that it is not challenging at all. Homework assignments are just a repeat of the lectures and take less than an hour if you took notes on the lectures. In other words, there is no value in the homework assignments.

The first two courses in this specialization were awesome. We did real life examples for homework assignments and through research you learned more than you had asked for. It was perfect.

Even in lectures, there is nothing 'applied' about this course. The professor just covers the content with no real-life examples. Very mundane and unexciting.

Also, why not talk about multi-label classification? Professor takes a real example with multiple labels (handwritten digits), makes it a binary class and then proceeds to explain it... Thanks.

My recommendation would be to restructure the homework assignments. Instead of having 7 questions that spoon-fed you the solution of a primitive problem, ask us to do some Kaggle challenges, or give us a topic that we go out and solve, do some peer-reviewed assignment. Lastly, if you don't have time or don't want to explain important concepts like pipeline, nested cross validation, and multi-label classification, add them as resources.

I am NOT confident in my ability to solve machine learning problems in Python from this course, nor is this course worth recommending.

There are various codes error in the modules. Test and Train datasets does not exist in few of them. Kindly solve such issues ASAP.

This is an excellent course. The programming exercises can be solved only when you get the basics right. Else, you will need to revisit the course material. Also, the forums are pretty interactive.

I know It is a hard subject to teach, but many ways to improve. Students could have been able to understand those concepts much better by using common or popular topics for assignment cases or practices. Without context, many useful concepts taught are forgotten right away. Lecturer should have not explain those concepts by simply talking to the camera without illustration or vivid examples. Often the way lecturer speaks and teaches is quite boring.

A lot of bugs in Assignments. Instead of learning ML need to go though forums/code to fix simple bugs like files locations readonly/ etc

This could the single most interesting course amongst all the 5 courses in this specialization. It made Machine Learning easy to interpret and fun to explore for beginners. The assignments are very thorough, though with some autograder issues. I strongly recommend anyone who's interested in ML to take this introductory course to again some knowledge in the different methods and applications of ML in various fields.

I just completed the third course (Applied Machine Learning course) over the last 7 days.

Good:

The course syllabus is quite well designed for an applied intro ML course

Assignments are nice & force you to think; you cannot simply watch the lectures & complete them straightaway; which is good in my opinion.

Needs to Improve:

The lectures are atrociously boring. The professor seems to be reading out from a teleprompter in a flat pitch.

There are parts where the intuition behind the concepts are well explained and others where you are left staring at stars and better off learning from other sources over the net.

The course seems to have been all but abandoned. Common mistakes in the assignment setup & lecture recordings have not been corrected since the course was first offered 2.5 years ago. The discussion forums keep getting spammed on similarly asked questions which can be easily solved by correcting the assignment errors and providing a few clearer comments/instructions. Week 3 lectures definitely need to be re-recorded as there is a correction prompt on every video. There is one 'Mentor' who helps out as a volunteer. No one else to moderate the forums.

The course pace is quite uneven and patchy. Week 2 is extremely heavy while week 1 super light. Week 3 is good but week 4 feels half done/rushed. Seems like there is an arbitrary administrative requirement to do a four week course from UMich.

All in all, I did not come away impressed & elated from the course. I did expect much better from my Alma Mater.

Although the course taught me a lot on the importance of parameter tuning and data leakage, I found that often times it was too technical and did not provide the information I was looking for. I found myself continuously referring to notes from other ML courses during the length of this course. In addition, the video errors and challenges with the auto grader were very frustrating.

great experience and learning lots of technique to apply on real world data, and get important and insightful information from raw data. motivated to proceed further in this domain and course as well.

Decent material and I appreciate the amount of hard work that went into building the course. However, the course should really be titled "Evaluating Classification Methods", as that is pretty much the focus of the entire class. The lectures (especially in Week 2) were SOOOOOO long and very hard to absorb, that even double-speed didn't help. In education, less is more. I would compare this course to the reading of a textbook. There was very little focus on making sense of the code and solving real-world problems and far too much emphasis on shotgunning (what felt like) every single classification technique known to man and trivializing pros and cons of each method. To make matters even more strange, PCA and other useful methods were pushed into "optional". This course should really be a two-part course, especially since the claim is that the course requires 18 hours of time. Sure, type in the code just as the professor does and you get the right answer, but meaning is lost of if you are to adhere to the timeline. If I didn't know more about machine learning and this class had been the first one I had taken, I couldn't run fast enough from the pursuit of a career in this field. Data analysis is intriguing and the methods are varied and fascinating. For me personally, this class was a let-down. Again, I recognize the course was hard work; I am merely stating my personal sentiments.

I would not recommend this course except for week 1. According to some forum posts, not only is this course a bit outdated (needs fixes to many parts and they haven't fixed them for at least 2 years), seemingly small but crucial parts of the assignments are not covered in the lecture videos which were very frustrating and time wasting to figure out (4 week course ended up being 10 weeks for me with a result of 93,4% final grade). Combing through the videos turned out to be futile and of course, relying on external sources such as stackoverflow was also not very helpful since the questions asked need to be extremely specific to the course. Even simply submitting the assignments were met with difficulties; and it's similar forum posts week after week.

While I appreciated the difficulty of the course, the poor design and structure of the course is evident with the number of correction pop-ups that come up every video. The professor misspoke countless times over the duration of the course, and there are several typos on the slides that need to be corrected! I feel like the videos were also much more bland than previous courses in this specialization.

Additionally, while feature selection and data cleaning are large components of the final project, they are not at all the focus of what is taught in the course! I think this course was trying to do too much all at once, and leaves you with a shallow understanding of several things instead of a good understanding of any specific thing.