Course is good to have some practice in spark and scala. However it seems to be long forsaken by staff and some issues with assignments require doing archaeology in the forums. Also it is quite unpleasant to see that while specialization emphasizes functional programming, some auxiliary code in assignments is written in the worst manner of imperative programming... Why use scala then?

If you know the subject, may be easy. If you don't know the subject, too dense and abstract with not enough practical examples/exercises before the assignments (specially for week4). Would have been better to break the videos into smaller ones, include more practical concrete examples/exercises during the videos (as very well done by Pr. Obersky during the 1st Course of the Specialization). There could have been room to break week4 into material for 2 weeks. Still, inresting.

Great introduction to spark. Fun assignments. Since it was the first ever session, there were quite a few kinks with the assignments. But the discussion forums rescued me any time I was stuck.

Let me start by saying that the content and the assingments are interesting and give good information. The reason I'm giving it a low score is because it's evident that this course has been abandoned by the staff (and they never payed much attention to it to begin with). First, the videos are blurry from the overlay of slides and recording, they also contain mistakes that are fixed while the instructor goes through them. They are videos and it looks like it's been 5-6 years since they were recorded, definitely time enough to fix those things. But my main problem are the assignments: They use really old versions of java, scala and spark, which leaves students trying to find workarounds to install and use these old versions. Sometime the code doesn't match what the assignment is showing. The answers in the forums are sporadic and, usually, not that helpful to begin with. All of this wouldn't bother me too much in a free course, but not only this is not a free course, it's also part of the specialization so, regardless of all the issues mentioned, students have to do it and pay for the certificate if they want to get the full specialization.

Grader was a bit counter productive, I have my points below;

second week k-means was a bit hard to grasp and it is not sitting on the core of this course.

last week assignment doesn't accept the answer when sorted in descending order and there is no clue it requires ascending order.

tests in the graders can be more descriptive on what is going on.

Apart from the assignment issues the flow of the course was beautiful, starting from the more primitive rdd's and going to more powerfull tools was very good in narration. Thank you for this course.

Very Nice and effective course. One of the best course i have done on Spark online. Many Thanks to the course instructor Heather Miller for creating a very detail and updated course on Spark.

a very basic and shallow introduction:

-spark dataframe usage scenarios and syntax poorly covered

-homeworks are too basic and do not prepare for real life scenarios

-lots of typos in the slides

-pdf slides are not searchable for terms

-how is this course rated 4.7 (as of 24-07-2020)?

A superb course taught by a superb lecturer. From a lecture stanpoint, my only recommendation is that additional lectures be provided to flesh out the coverage of Datasets. I found the assignments challenging and a good reflection on my understanding of the course material. I would recommend only the following tweaks:

Week 1 assignment: Either tightly specify the function that determines if a particular String is present in a large textstring OR provide the function in the assignment skeleton. There are too many ways to approach this function that are correct, but different than the one the grader used. This function is only incidental to the goal of applying one's knowledge of Spark to the assignment.

Week 4: Allocate additional lectures to flesh out the coverage of Datasets. The subject was not covered to the same level of detail that the other subjects in the course were given.

These are minor points.

Well done, Dr Heather Miller!

It's a great course with lots of insight into spark. At first it's confusing - first rdds; then dataframes and later datasets. But this is by no means the instructors' fault. Heather greatly describes in detail what goes on under the hood and why dataframes are way faster than simple rdds and why we have to give up some of our beloved typesafety of scala-land to let catalyst do its magic. Datasets try to bring some of it back.

I really enjoyed coding the assignments. Took me ~2hrs each. The tasks were fun and like those you would find in the wild.

I already applied some of the skills I learned here at work and successfully implemented a simple recommendation engine that will go to production next week.

Highly recommend this course!

Course is solid, useful concepts are thought. Assignments are interesting.

Points for improvement:

I wish this course was 2-3 courses focusing on topic with more assignments. If a concept is reviewed by the student is not challenged with tough questions the concept is not learned. People taking this course are already advanced enough to tackle difficult challenges, sometimes it appears that instructor assumes that students are entry level.

A note for online instructors: avoid jokes. They do not work online.

I think a specialization on applications of spark with scala covering AI, graph and text processing would be interesting. Overall thanks for the effort; pretty good.

The sessions where clearly explained and focused. Some of the exercises contained slightly confusing hints and information, but I'm sure those mistakes will be ironed out in future iterations. Thanks!

I think the course could be very useful for an engineer, and I recommend it if you are computer science and want to become a data engineer. However, as a data analyst, I don't recommend this course because it was much more technical and difficult than I needed to know for my current use at work. For this course, you need previous knowledge of Scala or at least know another programming language. Finally, the length of the course took me much longer than 4 weeks.

bad video quality, a lot of errors, very bad exercises without proper explanation

Very interesting course. Heather presents at just right level of abstraction to my taste and her presentation is very lively so it is easy to stay focused.

Formally, this is a course on a particular set of software tools and many such courses are not very useful unless one is to start applying learned skills immdeiately because the material starts getting obsolete the day the course is passed.

I believe that this course, however, teaches important priniples that outlive the particular toolkit, in particular the art of reasoning about algorithms on distributed collections, in particular, their performance. One would expect this to be a complex subject but somehow this course makes it feel simple -- which is a good indicator of high quality.

A good course with very nice and detailed explanations. I liked it. Two disadvantages that I found were: 1) Some theoretical knowledge of K-means and its proper usage from the previous course is required (and I didn't take previous courses. So I spent more time trying to solve this rather than the rest of the 2nd week's assignment), though it barely relates to Spark. 2) Starting from week 2 you need to write (well, not required, to be fair, you don't have to) tests by yourself. And I didn't get the idea behind it, to be honest. I think it would be much better if main tests were fully or partially provided so that I could check correctness of my code right in the IDE. Anyway, a great course, 10/10!

Pretty good overview. Could use some more fine-grained optimization exercises in Datasets. The first exam has some ambiguity and the use of imperative arrays in an otherwise functional specialization threw me off quite a bit. Some more specific unit test matching output would be much appreciated in the grader -- not the whole thing, but perhaps a different data set could be run against the student code with full expected and actual output than the one given to the students so that formatting issues could be addressed (especially on the first exam). Even as a person familiar with ML and Spark Processing, the formulation of that problem threw me for a loop.

I learned so much from this course! It was amazing how Dr. Miller used concepts that were meticulously built up in the earlier courses, such as evaluation strategy, functional collections, reactive programming, and associativity, to describe the core of Spark in only four units. As someone coming from more of a statistics background, I started this specialization only to learn Spark, so I wasn't always sure how relevant learning the more theoretical underpinnings of Scala would be. It turns out that it was pretty essential, while also just making me a better programmer. Looking forward to the capstone!

Great course over Spark. It shows the syntax, but more than this, it shows the problems, the caveats, the optimization and the architecture from a wide point of view.

The assignment were designed to focus right to the point: they managed all the configuration and initialization code of the project, leaving the student to fill only the most important part, with the resulto of having, at the end of the course, working projects that could be used as "cheat sheet" to remember all the material of the course.

Massimiliano.

I like the course. Very interesting and stimulating. Week 4 is a bit cumbersome, but it is due to a big amount of topics covered in it. I think that it could be better to have a DataSets and DataFrames brief overview in this course and study these topics in detail in separate advanced courses.

Assignments are designed to make us spend some time reading the docs. I think that it is good, as by doing so I have learned a bit more.

Many thanks to Prof. Heather Miller!

Amazing course! A lot of deep dives into technical details. Great excercises where you need to spend a lot of time and hence, will help you to improve an overall knowledge and gain more practice hours! Huge thanks to Prof. Heather Miller *Small comment of what to improve, from my perspective: the week4 task, timeusage, there were no any tests and also overall description was not totally clear, comparing to other weeks. Please improve this, if possible

Very nicely taught. Liked these "long" lectures of 15-20 min where the instructor gradually builds the material around the topic. Most other courses online, especially Udacity has frustratingly short videos where more than half the time the student has to keep clicking to the next video. Material coverage is sufficiently wide as well and the curriculum is freshly designed which is very important in this field.

The course presented the concepts in a clear and exemplified way. Tough, it was a bit too verbose.

The exercises were not too hard and thorough enough to cover the concepts presented in class.

I wished there were more exercises as Spark's API is huge and sometimes a bit tedious to use thus being hard to quickly get confident. I hope that the capstone project will help for that.

The course covers the important concepts and explains them in detail.

Homework tasks really make you think, revisit the lectures and read documentation, until you get it right, and that all deepens understanding of the material.

I also obtained some clues for the future: the lectures provided me with context, which helps formulate questions when searching for answers elsewhere.

This is a great course on the intricacies of Apache Spark. It is not a general Big Data course, neither is it an easy one. Doing the programming assignments properly requires reading a lot trough the Spark documentation, which I personally liked as part of the challenge, but beware if you are not that type of person or aim at finishing the course as quickly as possible.

Excellent course. I had some knowledge about Spark before but this course helped me to leverage my skills from pure theory to something practical. Now I have the confidence I can use spark to solve real life problems. This specialization as a whole is totally changing my career prospect. Thank you EPFL, Dr. Miller and Coursera for providing such opportunity for me.