- 5 reviews
- 3 completed
I found that the concepts were presented with little perspective and context, which made it difficult to absorb.
The course focuses on making plots. In my opinion there is too much emphasis on this. This is something that you could pick up online once (and if) you need to plot something in a specific way. In the contrary, the parts on clustering and SVD/PCA are treated very quickly and without a real explanation on the underlying concepts.
Weak. Poor English. Superficial yet lengthy (not straight to the point). No coherent structure (why spend 2 weeks on Machine Learning while there is another course from the program that focuses on Machine Learning). Very disappointing from that university - I did not persist for more than 2 weeks because I got the same frustrations from their Xseries of last year (Data Science and Analytics in Context). Have completed many online courses at a higher level in terms of maths and machine learning. A shame.
This course is very weak. A series of unrelated videos on broadband specifications in a monotonous format. I tried hard to complete it because I had validated the 2 previous courses of the series (which I found good and in line with the specialization's goal) but this one was a real disaster. I put some constructive comments on the forum but they got deleted.
yes the subject was very interesting and I did like the course. BUT, I do have a few remarks on my personal experience while taking this course. - I found the lecturespretty shallow. Except one week that focused on Spark, most lectures were not that useful or instructive imho. - The labs took me too long to complete, not because they were actually difficult, but more because they included lenghty texts to read, various blocks of code to review in order to remember what variable was what. And all this using the bulky iPython notebook format. What's more, for the most important questions, one often had to look for tips on the forums or search through the pyspark API manual because the concepts had not been properly taught during the lecture. Then finally once you have passed the test it is hard to tell if your solution elegant/efficient because of the lack of content regarding Spark programming in the lectures, and the absence of solutions once the deadline has passed. That was probably the most irritating to me. - Might be a detail but the deadline at 00:00 UTC was not clear (beginning or end of the day ?). Also I don't understand the point of a grace period if it is automatically provided to everybody. Better set one clear deadline than 2 unclear ones. - Finally, I found that the few mistakes in the material that appeared here and there were precisely on the more technical/difficult subjects (matrix factorization for collaborative filtering, joins etc...). And the usually responsive staff on the forums was suddenly silent on those. I still give a good rating because I do appreciate the effort put in by the teaching team (especially regarding the long labs), but I do think it could have been a much better experience, providing better learning in less time. As a comparison, I found that Machine Learning from Stanford was above this one, whether it was in terms of theoretical content (for example to understand how recommender systems work), or in terms of code optimization (how to leverage parallelism/efficient libraries) as well as on the presentation side (lecturer talking to his audience, not reading slides). Looking forward to part 2 now, Scalable Machine Learning.