It is no exaggeration to say it took me longer to complete this course than the first 3 courses in the specialty and the time was utterly wasted. I wouldn't object if I felt like I was learning new skills but it is mostly battling a poorly constructed course, with terrible assignments, a broken autograder and a Professor who is utterly disinterested in the education of his students.

When considering this course we need to separate the subject (which is fascinating) and the tools (which seem quite powerful) from the course itself. I had really high hopes during the lectures in week 1, where the videos are stronger and close to a well taught university lecture than others in the specialisation. However the assignments and the autograder issues are too great to ignore! Assignments are poorly worded (in one question it is literally trial and error) and the autograder often breaks. There are cases of people spending 10+ hours on work getting incredibly frustrated by the lack of feedback to find out the solutions were correct and the autograder was playing up.

Very painful going through this course although i have quite well coped with course 1-3.

But this course seems lack of systematic structure of building the knowledge, it just walked through the topics quickly and extensively. I had to spend a few hours to learn about the whole structure of text mining to build in-depth knowledge, more than 20 hours to watch the online nltk & genism tutorials cause i m new to text mining & nlp.

just hope the course can simplify the complicated topics such as where we are in the whole process, what's it, why we need it, working theory, coding, how we use these parameters, etc. to make life easier.

I would see autograder and unspecific instructions ruin this course.....Sometimes you know how to get the answer and the answer looks just right! but you still cannot get passed! I would not be taking this course if it was not part of this Specialization........ Improvements need to be made!

Instructor does not explain concepts, just superficially goes through subjects.

Some lectures lack coherence between subjects. you wouldn't know what is the relation between topics.

But it introduces some basic stuff which worth knowing anyway.

Instructor was poor. Inadequate coverage of the material in the lectures, some questions not clear as to what was expected. You can do better reading a book on this subject on your own.

This is almost a waste of my time. The structure can be clearer and the connection to Python is outdated. The assignments are poorly designed. The instruction is not effective.

I really wanted to like this course, and there were some redeeming features, but overall I'm unable to recommend it in its current state. IMO, the lectures were at much too high a level while the programming assignments were very detailed with vague instructions and little guidance. There was no link between what was discussed in class and how the fine details of the assignments were to be understood. In addition, the course was published with errors in the auto-grader and no resources in the Resources link (not even slide decks from the lectures, so to review material you were forced to re-visit all the recorded lectures which was very inefficient). My recommendation to Coursera and the Univ of Michigan is to completely re-do the course, doubling the number of lectures to provide not just the broad overview of the topics, but also some detailed descriptions of recommended ways to implement what was discussed. I would also recommend using Professor Andrew Ngs Machine Learning course as a guide for how to create great programming assignments, with detailed PDFs (typically 5-6 pages) describing what is to be done AND WHY (linking back to the lectures) and "telling a story" that is cohesive and leads the student to create something end-to-end (in small steps) that does something amazing by the end. The programming assignments in this course seemed, in contrast, to be a shotgun blast of "do this", "create this", "make this happen" with little context of how the small pieces fit together or what the overall goal of the assignment is to accomplish -- and at the end, a feeling of "I passed the autograder's expections, but have no idea what I've really done or why". There were so many great things that could have been done with the Text Mining topic, and this course touched on just a few in a very haphazard way that simply left me confused and wondering why I spent so much time to learn so little.

The instructor provided very low quality material.

I am an experienced online course learner, both with MOOC's and online courses through accredited universities. Unfortunately, in it's current form, this has been one of the worst classes I have ever taken. While it does have some interesting content, the delivery is sometimes wandering and more of a high level overview than a concrete, here's-how-you-do-it, practical class. The assignments also suffer from ambiguity and sometimes outright forgetting of explicit instructions. Moreover, workbook-type examples are often lacking. Although I'm very disappointed in the execution of this class, there is potential if these problems are addressed.

As an aside, after completing this class, I find it hard to believe that almost half the reviewers gave this class five stars. There are some fundamental problems here, and I almost gave up completing the rest of the series because of this one course.

.

This course makes you give up on data science and MOOCs.

Seriously, the content is poorly presented he keeps on speaking , telling 2-3 lines about a function and so on.

I highly recommend stay away from this pathetic specialization.

Curriculum is valuable but the course quality isn't on par with the other Applied Data Science using Python courses by University of Michigan. Week 4 assignment doesn't do enough to bring all the previous topics together in a realistic application. Week 3 lectures and notebook have teach the use of a scoring function wrongly - an issue addressed in forum threads for months but no edits to the video lectures and notebook have been made as of yet.

I have taken and passed all the first four courses in this specialization, and very much liked the first three courses. But the quality of this course on text mining is far below the average level of the first three. Go find some other courses if you want to learn text mining with Python.

There are too many areas of flaws in this course. I am only highlighting the top 5 below:

1. lacks good connection throughout the course content. This problem exists almost everywhere, both from slide to slide within a video and from video to video. Many times you would have questions in your head like “why is he talking about this?” or “what is this?”

2. use example just for the purpose of showing examples. Don’t really explain the point it is supposed to explain. In many times the examples do not provide clarity, but raise more confusion instead.

3. assignment tasks either too simple, or remotely related to what is introduced in the course. The worst case is assignment in week 4, where the assignment is so poorly constructed. You have to spent days to figure out the right answer. They call it “debug”, but there is nothing wrong with my code. I would say it is more of a process to “try to figure out what the instructor is asking for”.

4. talks too much about the theoretical things, not very good introduction of using python. Even when python code is demonstrated, it is almost always in a very abstract way. This is significantly different from the first three courses, and very annoying. You would need to spend about the same amount of time googling how the packages work as I have never took the course.

5. Repetition of content already introduced in previous courses, i.e., machine learning basics.

The most discouraging course in specialization.

Instructions in programming assignments are misleading or poorly worded. This is an issue with every module of this specialization but Text Mining has been spectacularily bad. You need to spend hours browsing the discussion group just to figure out what is expected. Mentors are doing a great job explaining in the forum, but there is no feedback loop - the instructions are never corrected. Sometimes you see a forum post about a misleading or simply wrong instruction, that is dated 6 months ago, and the instruction still hasn't been corrected. It's like no-one cares. I feel like 70% of the time I spent on this course wasn't learning Text Mining, it was dealing with ambiguous instructions or autograder issues.

I finished this course because I already finished 3 out of 5 courses in the total data science specialization. If you're just doing this course, I wouldn't recommend it. It's very heavy on theory, and the practical elements of Python are only touched upon slightly. Expect to spend a lot of time googling the answer to the weekly assignments, and reading through the forums of the course to find which slight edit you'll have to make to make it work. Oh, and the course instructors/teaching assistants are nowhere to be seen in the forum. There's been errors in the course itself and in the auto-grader that were reported 3 years ago that still aren't fixed.

Unclear assignment instructions, buggy autograder, and no instructor help.

Lectures are very good with a perfect explanation. More than lectures I liked the assignment questions. They are worth doing. You will get to know the basic foundation of text mining. :-)

Initially the lectures started fine. But by week 2, there is a big gap between the level of lecturers/material which are too superficial and the assignment which are very detailed. 90% of the time doing the assignments consist of looking up the forums or stackoverflow. The autograder is also severely outdated, never been updated for the past 3 years since the start of the course. Week 3 itself the autograder requires some "wrong answer" to pass, and this has never been updated. The mentors in the forum especially Uwe is helpful, but he's only patching the leaks by providing guides on passing the autograder. I'm only taking this course to finish the specialization, but I would not recommend this course at all, especially since it's paid and I feel it's not worth the price for the outdated content.

The first three weeks are fairly reasonable, but the last week goes over topics much too quickly with little explanation of HOW to apply the various approaches and what the models are actually doing. There's additionally no notebook for the last week. The lectures use Python 2 but the notebook requires python 3, leading to confusion. All 4 assignments are poorly worded in such a way that it's impossible to pass them without using the discussion forums. The material is interesting and useful, but the class is extremely frustrating.

Unlike the first 3 courses of this specialization, I'm very disappointed with this course.

It didn't give us a good feel of the technology, leaving a lot of blank spaces in the subject.

A lot of important subjects were just mentioned briefly, without training examples to make it clear, and those concepts were asked in assignments.

In Week 3 there is a concept that it's been taught wrongly, and everyone is pointing this out on foruns. But, even after almost FOUR years, this hasn't been correct.

We have to thank the mentors, those are the ones that are really putting the work patching those errors.

In week 4 the course didn't provide a workbook with examples, thanks for the mentors, they got us one.

Really disappointed with the overall lack of attention from the authors with this course. Especially after 3 great first courses in this specialization.

I hope that the author of this course is fired. I had hoped for more review and exercises on topics such as sentiment and topic analysis, things that are PRACTICAL. Instead the focus was on things like document path similarity, part-of-speech tagging, and counting words in spam vs. ham documents. Adding to the frustration is the autograder, and the arcane formats for submitting answers. Wish that I could give this less than 1 star, it definitely earned it.

I am a Data Engineer with a degree in Computer Science who wanted to learn more about Natural Language Processing for a small project I wanted to build. I had no prior knowledge of NLP other than some regular expression work from college and a basic knowledge of what tokenizing, tagging and classification were at a high level. This course was a great introduction into the field and has given me a solid applicable foundation to continue my education. I wanted something that was light in theory and heavier in application and this course hit a great balance. Contrary to many of the other reviews, I didn't have a problem with the autograder, most of the time I got an answer incorrect was due to not reading the question carefully enough. The assignments were great in my opinion and actually helped drive home the points made in the lectures. I recommend this class to anyone who wants to get their feet wet in the subject.

This course give the basic idea in each module existed in text and natural language processing kits. A lot more for self-explore, but this will intrigue to begin sooner and learn wider.

The professor was great in explaining the concept, but there was barely code alongs or notebooks with the code to show the outputs of the codes. In the first week, the code and outputs were provided, but as the course progresses, they are not provided anymore. I would have really helped to actually see the outputs and follow along with a jupyter notebook code.

However the reason why I am giving this course 2 stars is because of the assignments. The assignments were really a pain in the neck. Yes, the assignments help you learn, but it gets really frustrating when the autograder doesn't work properly. Most of the time the autograder grades incorrectly. In week 3, some of the solutions were marked incorrect when the solutions are actually correct. Instead of accepting the correct solutions, the autograder accepts the incorrect solutions as "correct" solutions. The autograder doesn't give much feedback on why the solution is incorrect.

In addition to the autograder, the assignments themselves don't guide you. The were so many additional things we had to do to pass the assignments. There were so many hidden tests that we were not aware of. And the assignments don't help guide us on how to pass those hidden tests. Week 1 and Week 3 assignments were so frustrating because of these issues.

Because of all this, I missed the deadlines for 2 assignments, so I had to push the deadlines back. When I did that, everything reseted. I had to redo all my quizzes and assignments. 18 hours of code was gone.

This course is close to a list of things you should know about for text mining, but provides little in the way of examples or detailed instruction. The fourth week doesn't even include a jupyter notebook with working code. The instructions for the assignments are hard to understand, misleading, or simply incomplete. The autograder during the third week is broken and won't accept answers that are correct but are of the wrong data type or which got rounded to 10e-11 as a result of the order of operations in the code. The mentor is MIA. Bad data science practices (like not scaling data) are tacitly encouraged ("just follow the instructions and don't add steps" says the Mentor) and sometimes you have to pass the wrong numbers to a function in order for the autograder to work. If I am hiring, I will not take what a student learned from this into account, as it was likely very little or perhaps wrong. Unlike other Coursera courses, this is a 'you get what you pay for' experience.