Comentários e feedback de alunos de Guided Tour of Machine Learning in Finance da instituição New York University

Content of the class is really good but technology/support is deplorable (Had to wait 3 weeks before the assignments got fixed by the support staff)

Not very related to finance plus most of the tasks are easy to complete, but hard to understand what needs to be done.

There is always something to be fixed in the assignments... Great content and relevance though.

Great content, but the labs are difficult to understand and often unrelated with the content.

There should be more explanations of codes in the video courses. Thanks.

Proper structure is absent but a lot of potential inside the course.

some python notebook has bugs, wasting time for me to fix

It lacks information on how to proceed on NN coding.

プログラミング課題でのプログラムの仕様がいまいちはっきりしない。

D

I write this review as someone who came into this specialization with prior knowledge of ML and RL but not finance. For me there is more or less nothing new here. Only a few finance concepts sprinkled here and there. The lecture videos are good as a refresher to basic ML concepts but this is definitely not for someone with no prior knowledge of ML as the mathematics has not been dived into deep enough.

I​ had hoped that the assignments would be made in a way that guides you through the specifics of ML usage in the financial domain but they are very generic. The assignments and demos are written using outdated tensorflow code, they need to be updated. Moreover, for someone new to ML, completing these assignments would be next to impossible. The objectives are not clearly defined in the assignments and there is definitely not enough background covered here for someone to be able to jump over that hurdle without prior experience. Also there almost zero support from the course admins. Overall, not a very good course. The only positive is the instructor. Hopefully the other courses in the specialization are better than this.

The lectures that are present are useful. However, I feel like the course is broken with some of the videos missing, as the lecturer references topics/items from supposedly previous videos that were never mentioned (this occurs specifically in Week 4, where the section "Prediction of Earning per Share (EPS) with Scikit-learn and TensorFlow" only contains basic videos with an introduction to types of equity analysis and what fundamental analysis is, but there are no videos with actual Scikit-learn/Tensorflow examples).

The weekly quizzes are trivial - they just recycle the knowledge check questions from within the video, and as standalone questions often don't really make any sense. The programming assignments are very sparse on instructions or information of what is expected. So while students do get some hands-on experience implementing some things in sklearn and TensorFlow, for the majority of the time they're 'flying blind'.

This course is more of mathematical introduction to machine learning than actual practical machine learning tips and tricks course. Math is definitely crucial but the way it was conveyed was not really good. I would have provided a refresher week just in math to refresh the students before jumping into the mathematics in the course. In the notebooks, there is a lot that was missing. Because I was already familiar with the material and I used TensorFlow, Numpy, Sklearn and statsmodels before and built several models with them before, I was able to navigate through. But if I was a totally new student, I would have a very hard time going through those notebooks. A couple of good notes, Please try to summarize all the important equations into a PDF file either for the entire course or per week to be as a reference when needed.

The course has rather decent videos, but the actual quality of exercises dunk after the very first one. Several exercises lack vital information in order to be able to successfully complete these without resorting to guesswork, while other pure and blatantly contains errors such as resetting the random number generator when taking new batches. In addition the solutions are so airtight, that rounding errors on the smallest of decimals causes one to get zero points, while the solution in any normal circumstance would be looked at as perfectly viable. Finally the version of tensorflow used is now so old, that the documentation has been scrapped from tensorflows own webpage, resulting in certain unexpected results whenever one tries to scoure the 1.15.0 documentation for an answer to certain problems.

I gave up while working on week 4's homework of the first course of this specialization. The two main reasons that led me to do so are: (1) very little on finance engineering except reference to problem cases and recommended readings; and (2) homework quality is really inferior to other machine learning courses I took at Coursera. I recognize that my first observation may not apply to the remaining courses of this specialization, but it is definitely the case in course 1. In the end, I thought I was not learning enough to justify the time and effort. Lectures are OK but they could be improved a lot by adding more financial engineering elements.

I believe that the course needs to improve the assignment piece. Instructions throughout the coding exercises are very poor. I understand that this course is for people with an intermediate level of python and Machine Learning knowledge, however because it promises to teach the practical applications of ML, some guidance it's needed. Even pointing out to a book as a reference for the algorithm would be enough. I completed the DeepLearning Specialization on Coursera and the quality of the teaching was way much higher.

Great theory. And good level of mathematical and statistical knowledge required to understand the concepts. However, It seems as though a lot of the coding aspect is brushed over and there is not much information given on how tensorflow works. Also, it needs updating to tensorflow version 2.

The course material are presented sparsely despite my initial expectation which may be formed by Andrew Ng in his ML course. Anyway I believe it is a good roadmap for learners of ML in finance and also for me to find and I should be grateful of the Coursera.

The course approach very interesting subject. However, it has incomplete informations and guidance throughout chapeters. I've felt much more informed by the recommended literature: Hands-On Machine Learning with Scikit-Learn & TensorFlow, by Aurélien Géron.

The lectures are actually very good, but I think it would help tremendously if you can make the slides and sample Jupiter notebooks used in lecture available to us. It takes us a lot of time to recreate the notebooks just to play around with them.

The quality of the videos is bad, is hard to hear the lecturer. Also the programing assigments usually don't teach a lot, is usually write down two or three lines of code for a 4 part assignment.

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.

Please help me how can I submit my assignment , No submission script is active in my course as well as in my programming assignment . 6th august is my last date of my certified course .

Lectures are good, but assignments are half baked, under specified and half the grading has errors. I hope this improves for people that take (and pay for!) this in the future

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.