The lecture is actually good. The positive experience is totally ruined by the quality of programming assignments though. As someone put it on course forum - they seem as if someone built a poor implementation with odd design choices in rush, then deleted a couple of random lines and asked students to read his/her mind. Not sure if I'll continue the specialization now.

Do not take this course before you review week 2,3 and 4 coding assignments which are wholly disconnected and arbitrary guesswork assignments where your task is to fill in missing pieces of code without any guidance or support. In its current stage the course is inaccessible to all but most tenacious learners with significant python and scikit experience.

One of the worst courses I've taken on Coursera. These courses really need to be tested before put out for public consumption.

Not an introductory level course. If you are new to machine learning, I would suggest taking Andrew Ng's course.....However some materials in this course are somewhat deep and rewarding if you have already got the basis..

The programming assignment is somehow painful and literally no introduction and demonstration of tensorflow is provided..... You need to do the reading and search the forum to get help to do the assignment

Terrible. For the first time in long time I felt such abandoned. No support. Notebooks written sloppy with plenty of copy-paste and no fixing. Thought more of the lecturer as well but videos feel like he's just coming up with the material. Having strong mathematical background I felt that the lecturer is intentionally making simple things sound hard. I'm left with deep sense of wasted time. Leaving Coursera and never coming back.

interesting but big gap between lectures and coding assignments

I rate the lectures and the lecture material a 5; however, the exercises are poorly documented and prepared and there is zero presence on the Forums from any of the TA's. The exercises, Forum and lack of TA's I rate a 1. Thus the 3 rating.

Excellent course with some tech glitches that are being cleared up.

1) Outstanding lecturer in terms of both ML and Finance

2) Real substance to the course - e.g. I do ML in finance and have for some time, yet I found this "guided tour" to offer some real opportunities for thinking and working.

3) I think that compared to ML classes that use toy problems to illustrate ML algorithms, Prof Halperin sets up the problems so that students have to figure things out. This is an uncommon practice, and I welcome it, but not everybody will.

For example,there was an assignment involving censored regression that required students to actually do some research - like, searching google or Wikipedia to figure out the special characteristics of the regression problem being posed, and relate it back to the code. The kind of thing one might expect in a college course. This stands in contrast to spoon-fed projects and assignment that are common in MOOCs. This is unfortunately mistaken by many students for an accident (it did not help that there were some technical glitches with grading early on). It's still easy in terms of poblem-solving in contrast to many Quant MBA -tyype courses.

So, for people who want to get a Certificate that they know ML for Finance without doing much to earn it, this class may not be what they're looking for. Those who want to learn a bit, and do so under conditions intended to offer some features of real-world applications, will be rewarded.

1) I don't really understand who is the target audience for this course.

For those who already have experience with machine learning, there is very little new information related specifically to ML applications in finance, most of the course is just explaination of machine learning basics.

For those who are new machine learning, it is too brief and lacks explaination of practical aspects. I don't understand how someone with no ML experience is expected to do these buggy programming assigments with almost no guidance and little lecture materials explaining working with ML libraries.

If you are new to ML, there are many MUCH better courses available.

2) Programming assigments are terrible. There are critical bugs in code templates, bugs in evaluation, messy and unclean instructions. These problems are reported in forum discussions for months but still not fixed.

I would give this class zero stars if I could. It is a great topic and I had high expectations. The assignments are poorly worded, instructions are vague and that is putting it mildly. The material required to complete the assignments is mostly not covered in the lectures. I can't believe NYU gives its name to this jumbled mess. Buyer Beware!

The assignments are very bad. Some content are hard to understand what it wants me to do. So little instructions about the formula and model, on the contrary, it needs the EXACT SAME answer with the EXACT SAME process of the assignment wants to pass it.

The Quiz also very bad. ALL the questions are THE SAME AS the control questions in the videos.

Though the course has good content, I will not recommend anyone to take it.

So many technical issues in the grading system. Also, Assignments are not clearly explained

Good lectures

Irrelevant assignments

No help on forum

Don't take this as a paid course to pass

Just take this as an audit course

Very Difficult - Impossible to succeed without very strong prior experience. Would deserve more guidelines

The course is not mature enough. If someone wants to learn machine learning in finance with efficiency and practicality, he or she should consider other options instead of this specialization/course.

The assignments of the last week were poorly planned, almost impossible to understand.

I learned much and got good practice in Python and Tensorflow as well as good exposure to the literature. I was able to download the course materials from the course system and work out homework on my own system for which I was pleased. The automatic grading system worked without incident once I figured it out and did not crash on me. On the other hand, some of the homeworks were less than fully explained and/or motivated by the course material and did contain errors and omissions in the supplied code that I had to track down in order to get them correct. The feedback from the grader was of no use beyond stating whether the answer was correct, but this is pretty standard. The course was frustrating at times and I would recommend it only for students who are highly motivated, but for those who are, it is definitely worth the effort.

More or less this course is good and interesting. However, homework assignments were awful. It's unclear and it's very hard to understand what is asked and how it would be graded.

was good but expect alitle explanation on the finance stuff

Objectives of assignments are not clear. The instructions provided in assignments are not clear. Tensorflow should be taught extensively because most of the students are facing problems in same.

A more detailed introduction and guide to python for machine learning would have made this course one of the best out there

Very goo lectures, but assessment exercises are not well defined. Examples not shown in lectures. Not enough briefing for starting exercises. No active forum for discussion.

Good material but assignments explanation were too sparse and even expectation of material not covered in videos or readings (example is Tobit regression in week 4).

Lectures are mostly short review of the topic. So you should know topics beforehand or supplement it with readings. Problems are great, you cannot solve it unless you understand the concept properly, so that good point.

To much math in lectures, assignments are not coherent and complicated, im not sure that i need tensorflow from scratch to work with finance(Keras fits better)