Comentários e feedback de alunos de Exploratory Data Analysis for Machine Learning da instituição IBM

Not clear pre-requisites. Instructions far off from the learning objectives mentioned in the beginning which makes it difficult to catch up.

This particular course as many others in Coursera, provides minimum possible knowledge with the lowest level of course quality. I will elaborate my point as following;

1) Instructor does not even know the actual mathematical foundations of what he is presenting. He provides example notebooks supposedly process a particular data which does even not exist. I personally and very discretely provided my comments regarding his conceptual mistakes in his presentations without receiving yet any feedback or observing a change in course material.

2) The final projects, even though presenters make money out of this course, are evaluated by peers. With that in hand I have a PhD in Physics, but somehow a random course taker who did not even acquire 10% of my math and coding throughout his/her education is evaluating my final project. Moreover, this person does not even understand well what is written in my project and gives me some random grades. As a result, I don't even get a feedback at all about my grade and or details of his/her grading.

Now, let me put these together. Coursera was a go-to place back in time. Nowadays its quality is not even close to be called 'mediocre'. I had the belief that at least some information can be gained and somehow it was worth taking class(es) back in time. After this horrible and totally not valuable experience, I do not think Coursera is doing a notable or at least an average job. I also have no faith in the comments that you guys publish here from your course takers. I have no reasons to believe them. I would like to clearly indicate that I am neither planning to take another course from Coursera, nor I am planning to suggest anyone to take a course from Coursera in near future.

Really Poor Teaching. Concepts that were clear earlier was made unclear due to poor intuitive examples. Few concepts were taught really well. But especially around the Hypothesis Testing part, the quality dropped very steeply.

I feel like the instructor's inability to explain things in detail stems from the fact the he doesn't really understand it as well. feels like:

Boss: "hey I need you to present this tutorial"

Instructor: "Sure thing boss, I just need to read it right?"

Boss: "Yes, but you also need to pretend that you actually understand it"

Peer reviews are also filled with a bunch of trolls who will give you a grade of 0 just for the fun of it - this was the final nail for me. I cancelled my subscription.

ADVICE BEFORE YOU DO THIS COURSE -- Look at the assignment and choose a data set that you can work with. Try and replicate the techniques from the explanation videos on your data set as you go through the course and then you'll be pretty much have a completed assignment by the time you finish the videos.

A slight problem with this course is the hypothesis testing bit of the assignment. The problem could be as deep as the ocean. If you choose a data set that you know you can get a good binary test from you'll cut down your completion time without losing any valuable learning experience.

The concepts are not explained in details. The instructor seems to read from a transcript which may not be the best way of teaching. However, content is great and it can help build a strong foundation.

One jupyter notebook is not able to run because a dataset and a python module needed for running the notebook is not provided. Lots of classmates ask about help in the discussion forums, however, no TA or any help is provided.

Does not go into detail and explain how to really code for hypothesis testing

great course content overall. couple thoughts related to improvement opportunities: 1.could you consider sharing more python sample code for each section? These samples do not have to be talked through - just there available for students to download and keep. 2. I had trouble submitting my course assignment initially due to the confusing instructions on the webpage. The page said Additional Comment box was Optional but it turned out that one would still have to put in "No Additional Comments". Otherwise assignment could not be turned in. This was a frustrating experience that could be avoided for others if the webpage instruction was more clear and consistent.

It does provide useful information but not much. There is very less hands-on practice provided.

In my opinion this course is really bad, the content was not that good and honestly it is not up to the level of a Professional Certificate.

if you really make the exercises and the final assignment the course really contributes you to better understand Data Analysis

I leant EDA in the uni with R. This course teaches the same thing in python and adds some extra stuff, really good course.

A very good course if you take it seriously! Good practical tasks where you learn much!

Excellent presentation. Learnt quite a lot.

COOL COOL COOL

The Course is quite detailed and well explained regarding the techniques and fundamentals required for exploratory data analysis. Sometimes although I found the contents being spoken in the video hard to understand because of the flow and the accent, but then reading the subtitles helped. Also, one suggestion would be to provide a presentation or some pdf documents for the most commonly used Python commands for various libraries like Pandas etc. for data handling (starting from data reading, cleaning upto hypothesis testing and further). This is because to makes hand notes of all the commands from the demo videos takes quite some time.

All in all, big credits to the team for such a well prepared course material!

The course was really helpful in understanding basic ML concepts and the computational framework we can use for EDA.

But a lot of students had problems with ghost reviews where they received 0 points across the rubric. It took me two days to finally get my assignment graded properly and lost significant time in correcting the problem. Coursera should really do something about this issue.

The only reason that I do not give it 5 stars is because the website of coursera is not good enough to handle the peer review assignments at the end of the course.

I​t doesn't cover T-tests and f-Tests which are often used. Also could be better descriptive about some topics.

not very clear and not detailed. jumping through courses without teaching basics

As with every IBM course, they tell you "not to hard code" but every project/practical exercise from IBM is littered with hard code. To the point where the projects are unable to be completed, without the help from one or two forum posts from a random student who has spent the time to find a solution. This is a growing problem with IBM's courses. I've learned more from other students, finding workarounds for your mess, than I have from the actual course work. Also, the content for this course, and any examples of code, was produced in Jupyter Notebooks. You didn't even create content in your own IDE, IBM Watson Studio, which says everything a student needs to know about IBM products.

E​xcellent course . Covers all the necessary information for beginners. Although I noticed people from non-statistic backgorund have a lot of misunderstaning about hypothesis testing and p-values which is briefly talked about in the course. I would recommend bootstrapping for non-statistic background students ( https://moderndive.com/ - Although in 'R', still an excellent site that teaches about bootstrapping in very simple language for beginners. I highly recommend it for all non-stats students)

I​ have one more suggestion, it would be really nice, if the course can add some examples about usage of hypothesis testing in machine learning besides research purposes like A/B testing, binning of categorical features and so on.

Although I had done such data analysis elsewhere in Coursera, this I found very comprehensive and systematic. I wish the topic of statistical significance tests was covered in some detail based on real data, rather random data generated for the purpose. I feel this area should receive more attention from the designers of the course. Thanks for all efforts put in by the faculty and all support person in the background. Thanks a lot..

Peer Review needs some moderation, someone marked all zeros, for one of my assignments. We are doing Machine Learning clearly an algorithm for such can be made available. Overall a great Introduction and hands-on guidance towards the Tools and Statistics involved for various business applications in the real world.