Mentors need your help if proceeding in the right direction

Hey ,
I am currently on week 2 of ML specialisation course 2 Advanced learning algorithms , i am unable to get a true feel of what i have studied till now so i have started reading Aurelien Geron’s Hands on ML , (finished one chapter) , Now i am confused whether to keep reading Gerons book till i reach the point where andrew has taught , or finish the course first then focus on book or do both simultaneously?
Because afaik , gerons book contains hands on implementation for ML programs which is missing in andrew’s course (it’s more conceptual)
~Kay

All of Andrew’s courses contain demonstration labs and practice labs.

@Kavalanche Since I have this text here are my thoughts:

  1. Personally I don’t see why they can’t both be done at the same time ? For one, don’t be overwhelmed by the size of the Geron text-- Only like the first 300-350 pages at ‘traditional ML’ related. The rest of the text is all Deep Neural Nets, which corresponds here to the second DLS series.

  2. Though Geron takes a practical approach, IMHO having theory as well is super, super important. I mean at the end of the day, you want to make your own models, right (?) rather than copy from a book-- And without the theory you will find it very challenging to make that transition.

  3. Myself, I am really liking the Geron text so far-- But also because, elsewhere I’ve already covered my ML/Stats-- But I did so in R; So already familiar with that, for myself it is just answering the question: ‘How do I do this in Python’ ? He also covers data visualization and other topics I am not sure are in Prof. Ng’s course.

In any case, that’s my opinion.

Not in that sense , but it’s like suppose i want to approach a new problem from scratch , how can i do that ? Andrew’s course is constrained to some fixed examples so i said that.

thanks for your time @Nevermnd , so should
a) i cover geron till where i have covered in andrew’s course and then proceed further with the course
b) cover andrew’s course first , the three part series , then refer to geron while practicing and building stuff
c) proceed with andew’s course , cover geron for existing course topics simultaneously and take out time to cover the previous topics like reg , grad descent etc.? like 2 days course cover , 2 day book refer for existing course , rest 2-3 days previous stuff?

You can take the outlines of the methods from the practice labs, and apply them to different sets of data.

1 Like

@Kavalanche in the end it depends on your comfort level.

They are basically two sides of the same coin.

Granted for Prof. Ng’s course there are no strict time limits, personally I would choose option c) myself.

Also keep in mind Geron is almost completely applied, i.e. using libraries. So while the studies in Ng’s class may to you at first seem ‘less relevant’ at least you have a sense of what the library is doing, how it does it, and why it is useful.

Knowledge is no good if you can’t apply it and application isn’t sufficient if you don’t understand it.

Just my two cents.

Best of luck.

1 Like

Thnks man