This course covers a lot about the data pre-processing, and the tools available in Google Cloud to enable the gruelling tasks. Thanks very much for the lectures and training labs. Very informative.
It's a pretty interesting course, specially that's the only one that teaches featuring engineering with a focus on production issues, but it assumes some knowledge with apache beam, and dataflow.
創建者 Yasim K
•The tf.transform and Apache Beam concepts are not explained in simple ways.
Also the lab jumped from simple programs to complex programs.
創建者 Robert U
•The assessments do not actually require writing any code; you just execute the given code blocks. Little knowledge will be retained unless students actually write code and solve problems, even for the motivated ones who read through all the given code.
創建者 Stephen R
•A lot of the code, did not work.
創建者 Mike W
•The notebook based demos are unfortunately pretty useless as labs. All of these courses would be much improved with real labs that require the student to build the system.
創建者 Adrian H
•A lot of the labs need updating and revising and made more meaningful.
創建者 Sudesh A
•The videos are good and better than the last two courses in the specialization; however, the labs lack proper instructions and not that helpful. This course seems like more of an advertisement for Google Cloud Platform than feature engineering: details of engineering part is hardly covered in the course; more emphasis is on demonstrating on how to do it on GCP.
創建者 Martin A K
•Would appreciate more guidance on the exercises
創建者 Ian M
•Had a lot issues with the quiz grader.
創建者 Jakub B
•Huge improvement from previous version, the notebooks actually run and use recent Tensorflow version. Still, some parts are abysmal, like quizzes that don't teach anything other than memorizing some facts from lectures, and labs that have extremely complex examples but do not require any effort from the learner, and don't test any skills.
創建者 Pablo I F
•Very bad subtitles, a lot of errors, so for the non english speakers it becomes hard to follow the videos
創建者 Dhruv D
•Probably my least favourite in this series. Never really dives into proper pre-processing and feature engineering beyond 1 good lecture by Lak and instead tries to shoehorn Google DataFlow and other services where possible. Worth skipping if you're time constrained
創建者 Aniket D B
•This course has less focus on feature engineering and more focus on GCP
創建者 Ayush T
•This course and the next course of the specialization is the most important course of the specialization. The reason is that other course except the first course deals with the working of APIs which might change in the near future but the insight that this couse provide on some of the topics is really really important, which I've not seen much discussed. This course is definitely a must-do.
創建者 Richard M H J
•This course starts to bring together the first three courses to apply TensorFlow. I have been waiting for us to get to this point in the specialization. Perhaps the background of earlier courses helps understand the Google infrastructure to support real TensorFlow problems. Perhaps I'm just impatient. Anyway, this course hit on a lot of topics but it is improving my use of TF.
創建者 Sinan G
•The course provides an overview and details of a very varied, comprehensive, and advanced range of possibilities to do feature engineering. Because the software and API's presented have a lot of details you will have to work a lot more with the information provided to attain a "hands-on" feeling. However you get a good starting point and knowledge of the possibilities.
創建者 Giovanni S
•Great course. A bit more difficult than the other 3, because the topic is more complex. Once finished the course you'll get the big picture. It may take some time to digest all little details, but everything is very well explained in a more than exhaustive way. Teachers are also very engaging and never boring. Highly recommend to anyone interested in the topic!
創建者 丸瀬重雄
•Various enhancements in demonstrating a practical case in feature engineering, starting from ELT through training, evaluating, and lauching an ML engine, taught with a lot of enthusiasm. Recaps of relevant ideas in statistics, algebra and calculus we learned back in our school days (things that some of us "used to know") kind of helped.
創建者 Mario R
•This course should be mandatory for any ML practitioner. It teaches you that ML is not only about throwing whatever you want to (sort of) a model and expect to get reasonable results. It is about getting to know your problem and squeeze the data available.
創建者 Jafed E G
•I enjoy the lectures. The professor has a good speaking and teaching style which keeps me interested. Lots of concrete math examples which make it easier to understand. Very good slides which are well formulated and easy to understand
創建者 Iman R
•This course give you knowledge about how to optimized your machine learning model based on real world case. This course tell you about few tricks to optimized your ml model and tell you how and when to implement the tricks.
創建者 Russell H
•A lot of great material that I have not seen covered other ML courses so far. My only complaint is that there is way too much material for a single week. It felt like it should be spread over two weeks at least.
創建者 Gowthaman S
•This course covers a lot about the data pre-processing, and the tools available in Google Cloud to enable the gruelling tasks. Thanks very much for the lectures and training labs. Very informative.
創建者 Omar M A
•It's a pretty interesting course, specially that's the only one that teaches featuring engineering with a focus on production issues, but it assumes some knowledge with apache beam, and dataflow.
創建者 Balaji R
•i really like the effort taken in developing this course, the structure. Kudos to Laks for converting lots of statistical and coding language to very simple understandable english.
創建者 Ting-Shuo Y
•Feature engineering is important but less discussed compared to general ML or DNN. Feature cross is a new concept and yet very useful for dealing with large datasets.