學生對 密歇根大学 提供的 Introduction to Data Science in Python 的評價和反饋

very tough tutorial is easy but assignment are very tough.

assignments are tough. didn't expect too much high level.

The lectures were not enough for the assignments provided

Good content but too fast pace and confusing assigments.

Jupyter notebook for assignments do not work properly.

The questions are very hard and not covered in training

Dint teach anything used in the assignments properly

The problem is the lack of practical exercises

It's asking for money to continue the course

A bit more motivation needed in this course

Good material, let down by instructors.

Its too fast paced and less elaborated

Nothing new. I recalled what I knew

The lessons were too fast and dense

Theory is not related to exercises

Good materials, annoying grading.

Should have been more detailed

Better reading a pandas book.

it's not very good

Not much of a use.

Less interesting

hard assignments

Too fast paced

This course was a travesty. 1. The version of Pandas being taught is not the current version.... so good luck applying this anywhere OR searching for help. 2. The lecture material was wiffle bat level then the assignments were mack truck level 3. I am a professional developer, I know how to use stack overflow and pandas documentation to solve problems. I was looking for a fundamental grounding of the materials. 4. I do feel I came away with a basic understanding of using pandas and python, but that's because I spent about 100 hours looking up answers to every question on here. 5. The lecture is so superficial that you'd learn a python way to do something, then a pandas way to do the same thing, then another pandas way to do something, then that would be the starting block for the assignment that would use advanced concepts. As a result I know 9 ways to do something simple with no recommended best pattern or understanding of when to use one or the other--and they all kind of muddle together now, but then spent dozens of hours researching the actual answers to the questions. "This is the way I like to do xyz, because of this. There are 3 other ways you may see and I'll briefly show you them" would be great. 6. For how important it is, the distinction between methods that mutate data and methods that don't was pretty minimal. 7. The online exercise thing is worthless. It uses an old version of pandas AND there are certain code breaking idiosyncracies in the tool AND it considers a pandas INT wrong if it's looking for an INT but there's no requirement in the question and no discussion of how to transform these or if there's any reason to do so other than to make the autograder happy. Look in the forum, there's straight answers like "an upgrade broke this, so it is not expected to work" which is a bad experience if you spend a few hours trying to debug code before looking up the answer. IT shakes your faith in all the exercieses. 8. This may be a coursera thing, but I'm learning this for WORK, I need to be able to get stuff working on my local PC. I see the autograder makes things easier, but it's basically a similar but different API. I literally spent 1 hour converting my code so the online grader would run for every 2.5 hours of local coding I did. It's debilitating. 9. This is probably a coursera problem, but it's really difficult to find the question you asked in the forum. Since you can't get through most of this course without forum assistance, that hurts. 10. I feel like I got gas lighted. You cannot do this class without already knowing python. This is mentioned in one of the lectures after you've already signed up. He recommends the Python for Everybody course, but it is very unclear from the Course Description before you pay money. Here are quotes from it, tell me if you would expect this to teach you python: *"This course will introduce the learner to the basics of the python programming environment", *"including fundamental python programming techniques such as lambdas", *"SKILLS YOU WILL GAIN: Python Programming". Then the forum is peppered with answers that say "This is not a python programming class". So... SUMMARY: This was my first coursera experience, I was very much looking forward to it and it really shook my confidence in the site. I did learn how to work pandas, but would have done just as well with a list of problems and a google. The "16 hours to complete" took me over 2 weeks of full time work--roughly 100 hours--due to both this disconnect between the lecture and the assignments and to the difficulties transforming working local code in a modern version to a buggy online grading system working on an old version of python but with some patches that also render legacy forums only 80% useful as well. The lectures manage to be both superficial and confusing (because they take a superficial topic then jam 4 ways to do the same thing into 30 minutes). And despite the course description you do not learn an intro to python here, just to data science. I will be trying one more coursera course, basically because all the other reviews on here say this is an abnormally poorly run one, but if they're all like this I will return to pluralsight soon.

The course is a great course in terms of the knowledge and experience of the instructor and the helpfulness of the staff. I gave it 2 stars for two reasons.

1) The videos are deceptively short. In MOOC instructional design, you normally design short videos because the attention span of an online learner is tends to be much shorter than an in-person university student. However, in this course, even though the video is short, it is really 5-10 times longer because they speed over the equations and teaching so fast that you have to pause and replay and rewind and replay several times while trying it yourself. So the timings are not truly accurate. If you were actually teaching it in a classroom with actual students you would go much more slowly. In this case, I wish they were more honest with the times by actually typing the code in real time while teaching. I would have preferred a lecture to make it more digestible.

2) The hardest thing for me about the course was the fact that instead of practicing computational thinking within data science (decomposition, algorithmic thinking, etc.), I was really just searching on stack overflow for how to put it into python. It's poor instructional design to only teach somethings and expect students to complete assignments without giving them all the tools they will use in the assignment. It would be ok if it were an accidental mistake, but this seems to be purposeful. This happened not just in course assignments but sometimes even in mid-roll video-overlaid quizzes where the answer was something not explained or taught or shown. This was really strange to me and caused a huge amount of time to be spent searching online or trying mid-lecture problems to no avail. It cause all the timings of the course to be off (#1 caused the video timings to be severely off and #2 meant that the course assignment estimates were HUGELY miscalculated). Good instructional design would mean that the professor should show all the tools one could use to faithfully complete and achieve the assignment. For some reason that was avoided again and again in this course.

Great material though. I loved learning. I just wish it were better structured and supported and that I learned more about the work rather than just searching online for how to write something.