学生对 密歇根大学 提供的 Introduction to Data Science in Python 的评价和反馈

"Course" is very self-directed. There are very few exercises, and there is minimal feedback on those exercises. No attempt is made to reinforce learned knowledge over time (e.g. spaced repetition) nor is there any definitive list for what knowledge should be gleaned from this course. Emphasis is placed on the student's teaching themselves and learning through experience with no attempt to actually understand the learning process or rest the course design on a scientific basis. Standard filler course for an empty education system.

The course instructions are very limited comparing to some of the other Coursera courses I've taken. The video lectures and code examples are extremely high level and general, while the assignments are significantly more challenging than the content demonstrated in the lectures. I spent a significant amount of time reading the Discussion Forum and googling for the right codes to use, rather than actually applying any knowledge learned from the lectures. If I wanted this much self-learning, I wouldn't have taken a Coursera class.

I have to say, as a new data analyst, even though I have started my work. It is still difficult to complete the assignment. The assignment is far away from the lecture video, which makes the assignment painful and you may lose interests in the process. Even the assignments are not written in a understandable way, so you have to go to forum to clarify each question in the assignment. So please make the learning process me more enjoyable and easy for the starters. Data science is a great career and and I believe it is the future.

Overall I have to rate the course sub par, so 2 stars.

The lectures seem disconnected from the assignments. If you're not already proficient at Python you'll spend your time on stackoverflow.com looking for ways to solve the questions. The assignments took me a lot longer than scheduled.

I did learn a lot by doing the assignments (using stackoverflow). I credit the course with "forcing" me to do the assignments.

Sophie (staff) was the saving grace when I really got stuck. Without her I would possibly have thrown in the towel.

i do not recommend this course to those who has just started Python and want to have a glimpse of the data science capability of Python, i do recommend those should have learnt Corey Schafer's Youtube learning channel for pandas and matplotlib and then maybe start this course, esp. with its assignments. The assignments were rather difficult for beginners, since it does not provide any feedback on where you are actually not right or needs improvement. The course is overall challenging to follow and get all assignments done.

Course materials were too brief to work on the assignments. You have to do lots of your own homework to get through the assignments. It is almost one can do assignments without referring to any of the class materials. It is obvious course cannot explain all the material but they could have organize it better by making clear focusing the assignments. Also it would be much useful if they can add comment in the Jupyter codes. Assignment 4 was very poorly explained. It is not tough at all but the explanation made it worse.

Videos are fast-paced, material is limited (no real slides or extensive doc). Exercises are sometimes not clear in their statement. Activity on the forum compensates that. Exercises are however close to challenges you face in real life. Finally, the start date was delayed and there is no clear visibility on when the other modules are started. So, the course could be better given, provide more material and be better coordinated. I followed one Python course at Rice University which was by far better given.

Pretty bad experience, this course let me feel like listening a people read a book. Professor just simply go through the material with relatively high speed, the connection between each part is unorganized, the graduate student even goes through the material with higher speed. Assignment sometimes confusing and autograder add even more barriers on my path. Assistance answer question quickly in discussion form which I'm appreciating, however, can't compensate for the disappointed quality of the course.

The course lectures were subpar and didn't cover the material in the assignments. Googling for answers and checking stackoverflow for code is a great way to get tips, but it doesn't teach you the core material. I would have rather had links to supplemental readings that the materials were based on. I would not recommend this course as an introduction course unless you've had significant experience with Python data structures and know how to work with lists, dataframes, and iterate over those.

Really disappointed with this course to the point I'm quiting -in spite of having paid for the grade-. It reminded me of my days at the university where I found being a great scientist/engineer didn't mean you were a great teacher. And this is the real problem, even things that should be easy to grasp are difficult to understand -just the opposite of prof. Bill Boyd who makes learning a joy-. Because of this I would only recommend this course if you already have an intermediate level in Python.

The professors are very clear and the speed if good enough so that if you have a computer science degree you won't get too bored.

But, ohhhh my good the grading is terribly frustrating. If you've left uni and spend more than a couple of months in the corporate world you will find it terribly frustrating. You had to fiddle around for hours around data format and some other potential issues.

For interesting python problem with a nice layout and helpfull solutions, discovery go to checkio.org.

This course was extremey hard. The pacing was fast, it seemed like the instructor wanted to end the lecture as fast as possible, I had to rewatch a 5-7 min video three times with frequent pausing to just get the idea of what is going on. The assignments were really hard, especially final questions. I would suggest making longer lectures with slower pacing taht would cover all the material you need. I have just finished this course and feel frustrated instead of motivated to learn more.

i really liked the lectures and i learned a lot of new stuff but the autograder was very annoying to me. i spend hours on transferring my answers to the correct format. There were no templates or good describtions for the right format provided so you had to read through the forum and talk to the stuff alot to figure it out. This time (many many hours for me) had absolutely no learning-input and it was very very annoying. I doupted many times if i would finish the course...

I felt like the test-candidate for a new course. While the course-work and the presentations itsself were excellent, the "homework" was a real pain as very often the Auto-grader didn't work and it took me hours to get a project which was correct, to be accepted as the right solution. Cannot recommend it to be honest unless they improve their systems. Also it was said that the other course specialization start soon but it seems they don't and I paid for the full package....

The course assignment was very hard—maximum time I didn't have any clue what I was supposed to do. The course instructor was not up to my expectation. I didn't understand what he was trying to teach me. The pandas version in this course was ancient, which Showed exception most of the time with the functions. I wouldn't suggest anyone take this course if really wants to learn anything. For a newbie like me in Python, I had to go back and forth to Stackoverflow now and then.

I couldn't get through the class. The problem sets bore no relation to the lectures. I understand that the point of that is so that you teach yourself, which is more "real world" but in this type of work there are multiple ways to solve a problem, and if we never see the right (optimal) answer, then it's very possible that what we came up with on our own is a super inefficient way to do it. I think the course needs to get the lectures a little closer to the problem sets.

The videos were fast, but informative. The exercises were nothing like the videos, I had to accomplish several hours of self taught learning. If the certification did not look good on a resume or linkedin profile, I could've accomplished the same outcome with just hours of reading stack overflow, watching youtube videos, and skyping with other students in my masters cohort. Would like to see the exercises in the same realm of difficulty as the tutorials.

The lectures and explanations need more clarity and better instruction quality. The assignment questions were often nebulous. This resulted in lots of time wasted with the grader. The discussion forums were the best part of this class. I'd encourage the course instructors to take a look at the University of Washington's Machine Learning class. Stylistically, pedagogically and content-wise - that's a much, much better Data Science class.

The pace is far too fast. I can't see anyone managing with this unless they are already somewhat familiar with the content. Assignments and quizzes are extremely difficult. When learning it is nice to have at least some easier questions that allow you to practice implementing what has just been covered. In this course it felt like all questions required you to link together all aspects of what had been covered so far in complex ways.

The auto-grader in this class SUCKS. It runs an outdated version of Pandas from 2016. Some of the lecture content also includes deprecated functions. The course content and auto-grader both need to be updated to comply with Pandas v1.0.x. It is EXTREMELY frustrating to spend hours getting code to run properly on my machine, only to upload it to the course site and have the auto-grader tell me it's incorrect or unexecutable.

It does not matter if you have intermediate python level of experience. You should not go through the material so fast and not explain properly what the specific code does. I understand that we should seek outside resources if we don't understand something but thats not an excuse for providing fast answers to what a code does and not provide explanations to specific code. All I saw in this introduction was fast typing lol.

The course content, which includes tutorials and assignments, is exception for beginners to hone their skills in python data science stack (numpy, pandas and matplotlib).

However, the automated grading process, I find, is ambiguous. Here's why:

First, we are not provided with a snapshot of expected output.

Second, the feedback doesn't support debugging; as a result, test taker will make clueless trial & error attempts

I have previous programming experience, sill I find the course is very challenging for the following reasons:

- course videos are very short (40mins per week), partly related to the required assignments, not covering all topics

- you need to spend most of your time (for me 10-15 hours per week) googling and reading Stackoverflow, and you might pick up a solution which is not appropiate for your current issue

The assignments were tough and they really tested your knowledge of python. That being said, the lectures were absolutely useless. I spent 90% of my time trying to figure out python through stack overflow. The jupyter platform is also very frustrating. For whatever reason, error messages kept coming up as I was trying to complete the assignments. At one point, I lost all my work because of this error.

Lectures aren't helpful. Assignments are challenging which is great but there is a difference between challenging students to make them learn more or to challenge them to such extent without providing enough resources that they go weary of the material. It is important to understand that students take courses generally to learn new material, not to revise what they have already learned or forgotten.