Chevron Left
返回到 Machine Learning Data Lifecycle in Production

學生對 提供的 Machine Learning Data Lifecycle in Production 的評價和反饋

423 個評分
76 條評論


In the second course of Machine Learning Engineering for Production Specialization, you will build data pipelines by gathering, cleaning, and validating datasets and assessing data quality; implement feature engineering, transformation, and selection with TensorFlow Extended and get the most predictive power out of your data; and establish the data lifecycle by leveraging data lineage and provenance metadata tools and follow data evolution with enterprise data schemas. Understanding machine learning and deep learning concepts is essential, but if you’re looking to build an effective AI career, you need production engineering capabilities as well. Machine learning engineering for production combines the foundational concepts of machine learning with the functional expertise of modern software development and engineering roles to help you develop production-ready skills. Week 1: Collecting, Labeling, and Validating data Week 2: Feature Engineering, Transformation, and Selection Week 3: Data Journey and Data Storage Week 4: Advanced Data Labeling Methods, Data Augmentation, and Preprocessing Different Data Types...




Interesting material. There are quite a lot of typos and many code snippets are directly from the tfx manual pages however the instructions provided and logic of the course is clear.



It is a very informative course. I learned a lot about data, metadata, schema and feature engineering, Also, Robert Crowe sir is a very good teacher.


51 - Machine Learning Data Lifecycle in Production 的 75 個評論(共 82 個)

創建者 EMO S L


Great content

創建者 Viktor K



創建者 Jennifer K


T​his is a very thorough introduction to data issues that arise when you go from proof-of-concept to project in production. It uses TensorFlow Extended components to illustrate workflow concepts, and the labs involve using these components in programming assignments. If you do all the ungraded labs, the programming assignments are quite easy.

創建者 Ivan P


T​o much emphasis on tensorflow, too few on underlying concepts, while we need it and alternative to TF. If the course was call "implementing <current course name> in TF" this would be fine, otherwise name is mileading. However, the course content is well structured and interesting, just 4 stars for a misleading name :)

創建者 Søren J A


This is a nice course. I specifically like the focus on data and implementation of trained models.

ML is much more than getting models trained , real life data, data quality control and continuous model maintenance is key to having succes with ML in a real setting.

創建者 Piero C


Overall, a good course. The lab activities have been planned extremely well.

S​ome concepts and definitions were a bit loose, and some quiz questions didn't actually reflect what was discussed in the lessons.

創建者 Jacob W


A comprehensive course. My only criticism is that in some videos the pacing is inconsistent where half the video is reviewing what will be covered and then it is very quick to go through the actual content.

創建者 Carlos A L P


I​ liked the intro to several techniques for feature engineering, validate anomalies between training and serving dataset but sometimes the labs didn't explain in details the steps implemented in the code

創建者 Wanda R


It's a new course so sometimes there are mistakes in the translations or there is something off in the assignment's grading, but the content is great. :)

創建者 Umberto S


Really practical course with good examples and a lot of materials on MLOps and examples on TFX to build and manage ML Pipelines.

創建者 Shayan H


The course is exciting. Lab and exercises are informative, but the answer to the quizzes are a little ambiguous.

創建者 Hassan K


It will be more interesting if unstructured data such as image, audio, ... is used more in the course.

創建者 Choo W


useful insights, but tfx implementation might be invasive towards exisiting mlops pipelines

創建者 Khaerul U


course material very good, but instructor very rare give example that make sense to me

創建者 Bharath P


excellent course. Nice to see how we can detect data drift and skew drift

創建者 Gonzalo A M


Sometimes this course is a little boring

創建者 Ryan C


The course has some usefulness but the videos are often sparsely filled with information and often repetative. In one section on feature engineering the course leader states that we probably know how to do this but we spend a significant amount of time recapping the basics... Most importantly the teacher is very difficult to listen to.

創建者 Hamad


V​ery theoretical course, more like reading a book. Less examples, more theory in the lectures. There should be examples in the video lectures on example datasets.

創建者 Daniel E


There is quite a bit of support coding that is required to perform many of the tasks in the final lab. It is what it is and I got through it.

創建者 Sagar D


Content is difficult to relate with, feels disconnected between modules and between different chapters.

創建者 Carlos C


It is too much Google oriented

創建者 Will N


I found this course very dissapointing, especially in comparison to the previous course that this expands on. Given the reputation of the speaker, I was expecting a higher standard.

To begin with, there is far too much focus on TensorFlow. The concepts in this course are important to know, however they are briefly introduced in the videos, which are followed by a TensorFlow coding lab. The key information is hidden behind what are called "programming assignments", which unfortunately are nothing more than regurgitating TensorFlow code. For week 3, the videos total 40 minutes, and for week 4, 31 minutes. This course would be improved by spending more time explaining the MLOps principles.

Many of the principles encountered in this course I have already been practising during my PhD, choosing to handcode basic pipelines to automate my ML analysis. I would say that while this course is useful, knowing how to automate a machine learning pipeline does not prepare you for the working world. Without understanding exactly what is being done when you run each TensorFlow command, you will not be able to understand what you are doing and this will limit the impact of the work produced. I have learned more through my own work than I have during this course.

There are alternatives to TensorFlow for automating ML pipelines and the demonstrations are not hidden behind a paywall. For that matter, there are a large number of videos online that demonstrate the use of TensorFlow.

創建者 Arturo M


I'm quite dissapointed by this continuation of the otherwise excellent Andrew Ng specialization.

I was expecting a course on frameworks and best practices for managing data in MLOps environments. Instead, this course is basically a commercial of Tensor Flow Extended, a MLOps framework by Google. Other tools often used in commercial applications (like cloud ML platforms) are not even mentioned.

It's true that the course does provide some tips, but they are often too general to be of practical use, specially for people with some experience in the field (e.g. "you need to validate your inputs").

I hope the next courses in the specialization are better.

創建者 Roberto N L


After the first wonderful course in this specialization, this one was quite a disappointment. While the topic is quite dense, the material covered in this course was very superficial and served only to sponsor TFX even though there are many other tools in this landscape. A more fitting name for this course would be "An introduction to TFX for the Machine Learning Data Lifecycle in Production:"

創建者 Nikki A


I was pretty disappointed in this course, particularly compared to the previous Andrew Ng course in this specialization. The last course was very informative and general, where as this one felt like a sales pitch for TFX. I learned very little, especially since my focus is on deep learning, not the shallow, tabular data that was discussed here.