Very frustrating. This course is a good Machine Learning overview, and light on programming. BUT the homework is based around an opensource library, TuriCreate - this is only available for Mac OS. Windows users will have a harder time with this course.

The workaround I found was to register for a student version of GraphLab (which the course previously used). I used an older version of Anaconda that I got from the GraphLab website, and modified the homework assignments to use GraphLab instead of TuriCreate

Nice overview of the specialization. Since it aims at showing the advanced and interesting things you will learn during the specialization, some of the practical sessions are way too advanced. Thus, for me felt more like a mechanical copying of what the instructors did rather than an actual assessment of what I understood. Also, since some of the applications are actually repeated at the beginning of the main courses, it feels like a repetition somehow when then you move to the specialized courses.

I really enjoyed the case study approach that's why the 3 stars but I'm not gibing it a 5 because some of the videos could just be skipped because half of them are the instructors laughing and the other half is some important info. Also it looks like they don't really care about the community because not all questions asked in the forums get answers. Finally, there are some clear mistakes in the Quizzes that haven't been resolved although many people have complained in the forums.

The course could have been marked by 5 stars if it weren't for the promotion of a commercial Python library developed by one of the speakers. There is no way a student could complete the course without having Python installed and a free licence acquired from dato.com.

Students should be able to use any programming languages and scientific libraries to do their homework and the subsequent courses of the "Machine Learning" specialisation are excellent examples of such approach.

Generally speaking the course is well taught and exposes the student to introductory level concepts in machine learning. My problem with the course is that it promotes the Turicreate machine learning library. For one thing, this library is not supported on windows. For another, the vast majority of machine learning professionals use an assortment of Tensorflow, Keras, PyTorch, etc... It would have been more proper to use one of those libraries.

I was unable to get graphlab running – had to use turicreate instead. Also, the most interesting part, deep features, came a bit "ex machina" – without a proper explanation how to create what was prepared. Also, I really miss the parts 5-6 of the specialization which look very interesting. The basics are already well covered at many places. If the parts 5-6 were existent, I would probably take the whole specialization. This way, I will pass.

I enjoyed the course and I will continue with the specialization. I am giving a 3-star rating as i) the lectures need to be updated with correct data or need to provide guidance as to when one should expect individual difference when following along with the notebook, ii) instructor / mentor response in the discussion forums is lacking, iii) graphlab is now an outdated tool as it is not commercially available.

It would be better for the course to focus on using scikit-learn for machine learning. The course focuses on using GraphLab (https://turi.com/download/academic.html), which is a commericial product, free for academic use. I'm doing this course for professional purposes and my preference is to gain familiarity with free/open source solutions that I will be later able to utilise in production environment.

To be honest, this course is not friendly to windows 10 users because it forces students to use the apple Inc's Turicreate instead of the most popular sklearn. Admittedly, windows 10 users can still install the Turicreate by WSL but not everyone wants to add a subsystem to their windows just for this course. Except for this, this course has a nice structure and the content is really practice-oriented.

I found the Course very interesting, well prepared from the Tutors and I liked the case study Approach since it provides actual examples where Machine Learning can be realized. I am interested to enroll in the second Course of the certificate to validate if it will go into more Details and Background regarding the build of the algorithms theoretically and in Python. I would like to thank both Tutor

While I am aware that this is an overview of the other courses in the specialization, I felt that the quizzes and programming exercises didn't really get into the actual topic. For example the recommender systems quiz and programming assignment have nothing about factorization except a single superficial question. The material is clear and the overview is nice, but the practical part let me down.

A nice ML overview that introduces many tools without going into detail on how they work. Pro: Loved the programming assignments, nice Jupyter notebooks. Con: found the constant hyping of the Capstone course (which got cancelled) frustrating. The GraphLabCreate software was neat to see and easy to use, but ultimately I preferred the more first principles approach of Andrew Ng.'s ML intro course.

This is a good introductory course, however the quiz questions and over dependence of graphlab are off putting. The instructors share good insights about the need and motivation for various ML techniques. I wish there was more support on the project using pandas and sklearn. Graphlab is immensely powerful, however not adopted in industry making it hard to apply the learning in real world.

Pretty high level overview. I guess it's necessary to give a roadmap for where the concentration leads, but I wonder if each lesson couldn't have been added in its respective module, or if at least the Foundations Module couldn't be shortened a little - or alternatively made a bit more challenging. I'm on the first real module now and the change in difficulty is quite significant.

This course is for you if you really don't know anything about Machine Learning and nothing about Python. If you do know something about it, look for a different course.

I learned the most from lesson 5 and 6 about recommenders and deep learning because I knew nothing about these subjects.

The programming exercises are disappointing: just cut and paste. I found this demotivating.

I wish they used open source tools for this. I will not be paying for a GraphLab account nor do I see myself using it in the future. I felt less inclined to strain over learning the API and was unused by the technologies. Other than that the course is pretty interesting as I was able to apply some cool data analysis using ML practices I've learned in other Coursera courses.

The course was good, and the instructors did a good job. There don't seem to be any mentors in the forums who are helping, and the library used for the exercises was changed from the one in the lectures. The specialization seems to have been abandoned before they published courses 5 and 6, so ignore every time they talk about how great the capstone project is going to be!!

Lectures are great. Unfortunately, i can' t install graphlab create on my windows 10 labtap.I wasted two whole day on it!!!!! I tried every methods google told me, all fail or with bugs. I think pandas and sklearn are far more user friendly.不建议大陆使用windows的朋友尝试安装graphlab create，标准安装方式即使用了VPS也网络链接失败，用anaconda安装的话，anaconda3可以安装，但是没有canvas功能，anaconda2各种奇怪报错。搞了两天失败，我还是用sklearn。

This course needs to be updated. Windows don't support TuriCreate or Graphlab. Because it works on python 2. But now python 3.8 has come and TuriCreate doesn't work in it either. So, I had to use Ubuntu in my virtual box to work on the assignments. The course wasn't bad. But if they update the course, it will be a great one for beginners in machine learning.

I think it was a good introductory course. However, I think it was too simple: assignments required no more than copy+paste from the lectures.

I understand the primary goal is to hook people up on how good graphlabs is, but I'd rather leaarn numpy, sklearn and other widely available tools. At least show both in the leactures. Please :)

This course presents an overview of different machine learning tools but I rather prefer starting from the second course were more specific competencies are given. I believe that in courses like this the contents are to sparse.

It would serve as an introduction. But the contents of week 4 and 5 are not even in the specialization!

The course teaches a a lot of information and explains everything from a beginners POV which is great. I only have 2 issues with this course, the use of proprietary software instead of all open source software and NO CONTINUED SUPPORT for about 3 years since the course has been out. Either update the specialization or cancel it.

Wanted a course on ML. Could not find anything except this. Therefore decided to take it. However not so sure Turi Create/Graphlab is the best option (had installing issues as a Python beginer). Was it worth to use it instead of Scikit learn? Also I would expect closer connection between quiz tasks and study material.

Good for start in machine learning concepts. Good because they use Jupyter Notebook an python (they use python 2, it would be better 3). But I don't like much the graphlab library for data frames. And also the quizzes are a bit difficult. You have no the whole information in the documentation to solve them.

The content is really good, well explained=> 5 stars, nothing to add.

Why the 3 stars then ? Graphlab. If you use the pydata stack, as it is said to be possible in every assignments, you get stuck on the quizz due to variations on data and the algorithm used, so you can't really get quite the same answers.