I give this course a 4-start rate; it was a very informative and well-structured course. The only downside of this course is that they used some proprietary library, which has changed significantly since they used it. However, they updated the documents and added some tips on achieving the same result with open-source modules such as Pandas and NumPy. Finally, week 5 wasn't as good as the rest of the program.

An excellent overview of regression techniques in Machine Learning, with a very well thought-out balance between explaining concepts, providing enough maths to support the concepts (even with some optional "deep dive" lessons). For those interested in the really technical details, I think this course is an excellent start to get a grip on the concepts before diving into formal proofs. Highly recommended.

Really enjoyed the course - I did well, but this is really in-depth material. I feel like it would be really difficult to implement an ML regression algorithm from scratch in a job material.

I would like to see more interactivity in the lectures (short-quizzes interspersed in the videos) in addition to the long programming assignments at the end of the course.

-Thomas

Ridge regression could have been explained better. The best explanation was for kNN; perhaps this could have been the first module; since it is so disjointed from the others anyway.

The main reason I am rating this lower than 5 is because the notebooks for the assignments were only Graphlab based. Please do consider also giving notebooks that use pandas/numpy only.

This was much more in-depth than the intro course in the sequence, which was exactly what I was hoping for.

I still think that it could stand to be more challenging. Perhaps the instructors might offer some optional, more challenging exercises. Or, maybe students could choose an alternate "challenge version" of the homework that contains fewer hints.

Wonderful lectures and good assignments. Very, very clear presentations.

Minor drawbacks: there's no assistance available for the assignments (which can be quite difficult). The quizzes require sometimes information that is not directly available in the video lectures.

Note that it takes more hrs per week than mentioned (but it's worthwhile!).

It is interesting to understand how the gradient descent and other optimization algorithms works but it took al lot of time. In my opinion, that time could be used with another examples and practicals applications and even programming practical algorithms. After all, the course is very good produced! I will recommend it! Congratulations!

Better than the first one, however I feel that some material was missing and that the last week was rushed. It is sad the don't talk about XGBoost and other recent algorithms for regression (Elasticnet and so on). I also think they should only focus on using Numpy/Scipy instead of closed source software they recommend.

It was a great learning, concepts are explained in so simple manner. Thank you.

On the exercise, i think it would be better to test concept than the actual result value as different people uses different python packages and sometime result may differ slightly which make it difficult to pass. Overall simply Great!

Good introduction to linear regression. The quizzes are moderately difficult, and can be somewhat time consuming. There is a lot of depth to this topic, and tho this course touches on a lot of it, it won't be enough to get a very thorough understanding. In general I'm really glad I took this course

The course was mostly well taught, and having the two programming assignments each week was useful. One so you could learn the technique, and another so you could learn the underlying algorithm. Some of the topics were occasionally light and too fast on the detail, but overall a very good course.

This course is better than the previous one is this specialization the video lectures very good and topics but programming assignments looks to designed to use sframe and graphlab so it has some problems but this time there is better supporting for sklearn and pandas so i recommend this course

The courses get better and have more assistance for those of us who can't / didn't use graph lab. It is still outdated as python 3 came out after the course was created. But did learn a lot of stuff. Module 4 was the most frustrating as you'll get the wrong answers if you use pandas/sklearn.

Good course that covers the intuitions and goes a bit deeper with the implementation. Main down point is the fact that they try to get you to use graph frames and the alternative pandas route is often a bit of an afterthought, occasionally leading to trouble/confusion with certain quizzes.

The course dealt with implementing the key regression algorithms from scratch as well as tuning the hyper parameters using cross validation and observing what the pros and cons of each algorithm was.The course explained the maths/stats knowledge well and I would recommend it.

Once again, what a simple way of presenting concepts of gradient descent, coordinate descent. One thing to improve however would be to present the results of the quizzes when they're validated, will all the right answers, along with explanations. Thanks once more!

Amazing course uncovers all the abstraction behind Regression . Presented all the mathematical algorithms in such a convenient manner that results in efficient coding. Provided Notebooks are very well designed and presented every coding method in an easy manner.

The videos are great, well-structured and introduce gradually the complexity. Unfortunately, the exercises requires the use of a specific library, instead of scikit-learn and numpy. Furthermore, they also required Python 2, while Python 3 is now widely used.

Excellent course overall. Only issue was the inaccurate instructions for those using sklearn instead of graphlab (with graphlab it's very easy to do, but a bit painful for those like me who like to write a program instead of doing in interactive mode only).

The design of the course and presentations are great. It was very useful for my career development and fun. But, I think that the material is outdated and need a major update, especially Python packages and codes. Also, the forums are not active anymore.

Some of the quizes and assignments had grading issues but I thought the curriculum was great. All of the optimization techniques learned to solve for regressions seem like they could be useful in other areas of machine learning and other applications.

In the end of the course it became a bit boring to work with the same dataset over and over again. But I learned a lot about such algorithms as LASSO, Ridge regression. I guess that overall there's a good balance between math, theory and practice.

Excellent lectures, evenly paced and nicely balanced between theory and practice, and mostly great quizzes and practice problems. However, the complete lack of instructor participation in the forums leaves many student questions unanswered.

Like all coursera courses, inadequate guidance in notebooks to prepare for tests. No feedback or guidance on instructions when you get stuck (e.g, write this function to do thie following, no way to help figure out how!)

The course was very helpful. It would have been better if the tutors explained how they have arrived at some values they used like initial weights, tolerance, threshold etc.., so that the learning is comprehensive.