Chevron Left
Retour à Big Data Analysis with Scala and Spark

Avis et commentaires pour d'étudiants pour Big Data Analysis with Scala and Spark par École polytechnique fédérale de Lausanne

4.7
étoiles
2,331 évaluations
476 avis

À propos du cours

Manipulating big data distributed over a cluster using functional concepts is rampant in industry, and is arguably one of the first widespread industrial uses of functional ideas. This is evidenced by the popularity of MapReduce and Hadoop, and most recently Apache Spark, a fast, in-memory distributed collections framework written in Scala. In this course, we'll see how the data parallel paradigm can be extended to the distributed case, using Spark throughout. We'll cover Spark's programming model in detail, being careful to understand how and when it differs from familiar programming models, like shared-memory parallel collections or sequential Scala collections. Through hands-on examples in Spark and Scala, we'll learn when important issues related to distribution like latency and network communication should be considered and how they can be addressed effectively for improved performance. Learning Outcomes. By the end of this course you will be able to: - read data from persistent storage and load it into Apache Spark, - manipulate data with Spark and Scala, - express algorithms for data analysis in a functional style, - recognize how to avoid shuffles and recomputation in Spark, Recommended background: You should have at least one year programming experience. Proficiency with Java or C# is ideal, but experience with other languages such as C/C++, Python, Javascript or Ruby is also sufficient. You should have some familiarity using the command line. This course is intended to be taken after Parallel Programming: https://www.coursera.org/learn/parprog1....

Meilleurs avis

BP

Nov 29, 2019

Excellent overview of Spark, including exercises that solidify what you learn during the lectures. The development environment setup tutorials were also very helpful, as I had not yet worked with sbt.

CC

Jun 08, 2017

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!

Filtrer par :

401 - 425 sur 459 Avis pour Big Data Analysis with Scala and Spark

par Manuel W

Apr 18, 2017

Would be better to have more and shorter exercises.

par Ruslan A

Aug 23, 2017

lectures don't correlate to practical assigment :(

par David G

Aug 25, 2017

Great course, but can be great idea have the ppts

par Yuan R

Jan 20, 2018

Great course that is very practical for the job.

par Guillermo G H

Jun 30, 2017

Great approach to learn about Spark in practice

par Michaël M P

Feb 05, 2019

Talk about how to set Scala version in Eclipse

par 林鼎棋

May 30, 2017

Great! But I want to know more about dataset!

par VeeraVenkataSatyanarayana M

Jun 04, 2017

Basics are covered in an effective way.

par Pavel O

Aug 12, 2017

Good final course for Scala learners.

par Lucas F

May 15, 2017

Great lectures and great content!

par Васеев Р В

Jun 24, 2018

I would like to learn some more.

par Hoon P

Apr 19, 2017

Learned Spark APIs, internals.

par Alberto P d P

May 12, 2017

Very good and concise course.

par Stéphane L

Oct 13, 2017

Very useful

par Srinivasu N

May 15, 2020

good

par Devaraja K R

Nov 14, 2018

Good

par Jim N

Apr 12, 2017

I understand that creating such a course is a tremendous amount of work. Let me congratulation the developers for all their hard work. I learned a lot. I'm glad I good the course, and I'm speechless that I finally finished it with a passing albeit imperfect score. This was the maiden voyage and it has some problems that should probably be corrected. That's my opinion.

There were generally two sets of challenges with each assignment exercise (1) undemanding the scala/big-data component, and (2) understand what's being asked. Sometimes the comment in the code are wrong or misleading. In particular in the 2nd set of exercises. The functions are very poorly specified, and there are no unit tests. there should be one or the other. Without a specification the user cannot write his own tests. In addition the 2nd exercise is intended to be do-able, but not efficiently until the user views week 3's videos. Because of this the text output is EXTREMELY verbose, making it impossible to read the error messages and understand them. When the user finally submits the assignment it fails and the grader gives misleading error messages. A failing unit test should at the very least explain what the test was, what the measured value was, and what was expected. This is especially true when the user is forbidden from looking at the actual testing code. An additional problem with exercise 2, is that the student is asked to calculate a particular percentage, but it is not explained what this is a percentage OF and there are several ways to interpret it. I didn't realize until wasting 2 weeks that there was an alternate interpretation which I could try.

In the 3rd exercise some of the comments in the code are wrong/misleading. Particularly with regard to classifiedColumns. comment #3 should read:* 3. other activities (leisure). These are the columns starting with “t02”, “t04”, “t06”, “t07”, “t08”, “t09”, “t10”, “t12”, “t13”, “t14”, “t15”, “t16” and “t18” BUT EXCLUDING those which are not part of the previous groups only. Otherwise the sentence is at best grammatically incorrect, and at worst misleading.

Another significant issue with this "Bit Data" course is that it assumes a user understands data base and can construct sql queries. Well, such background is not a prerequisite, but the student needs such information to successfully finish the exercises. I can understand the temptation by the course developers to include a section on SQLspark, as this is a very powerful set of spark libraries, but I believe it is beyond the prerequisites. I could not have finished the course if I hadn't had an SQL/data-base expert in the same office, who I could pose questions to.

I hope you find my comments useful.

par Harold O

Apr 16, 2017

This is a really excellent course. It deals with difficult content clearly, thoroughly and at a good pace. I would now be happy to use Spark in my work and feel I have a strong base to go on to further study. The exercises are well conceived covering fairly realistic use cases and are of about the right complexity. BUT, the feedback from the automatic grading really lets it down.

The auto grading tests are at too high a granularity, each test tests too many things. They give cryptic errors such as '<some random element> is empty'. You have no idea why the element is empty as the students are given a different dataset to the one tested on. Moreover, the failed tests are invariably nothing to do with learnt course content, but trivial rounding/formatting errors. The forums are a lifesaver, but I still spent a good two weekends working on bugs once I'd got the assignments 70% right. I would not recommend to a completist!

All the same, thanks to Heather for the wonderful work she's done putting it together and the excellent lectures she gives.

par Tom C

Apr 05, 2017

Good information, but a little rough around the edges, possibly due to being its first time out of the gate. Should be amplified with additional weeks of topics, e.g. on Spark Streaming, etc.

The Grader often returns opaque errors ("The Grader Failed"), when the actual problem was that the program ran out of memory or out of disk space. Grader can reject solutions, e.g. on the last assignment, if the programmer uses a different rounder implementation that returns subtly-different results, though the results print out as the same to the rounded degree of precision.

The Grader should be written to accept results that are "acceptably close" to the desired result, e.g. if the desired result is 34.1, to accept perhaps numbers between 34.099 and 34.101 -- perhaps consult with other course-runners for best practices here.

It might be helpful if the course provided a quick intro to running the assignments on DataBricks.

Good first effort.

par Aleksei M

Apr 19, 2017

The selection of content is very good. Yet, the course requires solid amount of polishing. I had a feeling that the real teaching started only in the 4th week, with lots of walking around the subject in the beginning. Worth to mention that in contrast to the first 3 courses in Scala specialisation this one is very heavy on visual aspect. You actually have to see the picture, as the voice channel just references it.

I can certainly recommend this course to people curious about data analysis. If you have had some experience and would like to try Spark or get better at it, then probably a good book can save your time.

par Sam Z

May 03, 2017

good: the concepts and on-hands skills taught by the course are good.

bad: the assignments. you cannot complete them just by following the course material, forcing you to waste quite a lot of time either: (1) learning from other sources; (2) looking for answers on the forum; or (3) brute forcing an answer till rage quitting :)

another bad point: the course is supposed to be focused on spark & big data analysis but it has 1-2 lectures (around 40-60 mins) pretty much devoted to showing some SQL. << this could be summarized to around 10-20 mins and/or give a link to quickly learn/try to some other source.

par Horia R

Apr 05, 2017

The course was interesting but it was clear that it is the first iteration. The course seems rushed, there are not very clear explanations on certain areas. Mostly, I had problems completing the assignments based on what I've learned during the lectures.

The previous courses had difficult assignments and you had to think about how you wanted to do something, here the problem was using the Spark API and understanding things which weren't explained in the lectures. Also, not all the areas of the assignments were described.

I am sure the course will become better in time.

par Allen S

Jul 11, 2017

It's a good course with still some room for improvement. Some lectures are too long and lack dynamism. The code of the assignments is surprisingly un-Scala-ish (var and for loops) for a Scala Specialisation course.

On the plus side we get a very good understanding of the basics of the Spark engine, the different APIs and the SQL module with hands on practice and a forum, all the benefits of a Coursera MOOC.

It's a good start for those who want to learn about Spark with Scala.

par Rob S

Oct 15, 2018

Lots of good knowledge in this course. Took a star away for the second exercise, which taught me more about how to pass the second exercise than anything useful. Took a second star away for not having test coverage that helps the students. You wind up submitting and waiting for the grader to run.

Yes - I know we can write our own tests, but in a graded course like this with sometimes unclear problems the tests can make all the difference in the world.

par Luis V

Oct 01, 2017

Good course but with many outdated concepts (mostly valid for Spark 1.x) and some pitfalls. Need many improvements, actualization and some reshaping in the distribution of the topics and sessions of the course. The topics left for the last week are some of the most important and central in current Spark 2.x and they include at least as many fundamental concepts as the rest of the course.