I'm now playing around with this new basic blog template from Next.js examples --> available on Next.js GitHub.

Luciano Lupo Notes.

My take on Machine Learning on GCP course

Cover Image for My take on Machine Learning on GCP course

After finishing this intro to Machine Learning on google cloud: https://www.coursera.org/account/accomplishments/specialization/VXUJ2GFL63K7

I think it is a good intro to BigQuery and Tensorflow and how to use them on GCP, also a good exuse to kee on learning about GCP it has real world exapmles and practical Jupyter notebooks to play with, also you can run it all in the platform for free so you don't need a powerfull GPU to runn al the things that are being taught in the course. google uses it for lots of its products ( like google photos and google translate ) and uses tensorflow along with a whole plethora of other tools but this course keep it on the basics.

First lab

First lab is about Framing an ML problem, it is a good way to start thinking about the problem you are trying to solve and how to approach it, it also gives you a good understanding of the data you are working with and how to explore it, what you need to predict and what you need to use to predict it.

Then continues with google services and basically self-propaganda:

ml google cloud image 1

continues with explanations about why data quantity and quailty is almost all the time better thatn complex models, most of the times ML problems are big data problems....

and the most important thing, the 5 phases of ML (according to google)

ml google cloud image 2

The next part is about Human Bias, metris and evaluation, it is a good part to understand how to evaluate your model and how to avoid bias. and all the rest is about using their python lab on the cloud( I think te most practical thing of the first course ),the interactive hands-on labs platform called Qwiklabs.

Best course is the 4th one( in my opinion... ):

Is the Feature Engineering course: https://www.coursera.org/learn/feature-engineering

feature engineering is often the longest and most difficult phase of building your ML project, and in this part of the course I've started with raw data and the point is to find relationships between data and features and also how to identify good features.

we use Feature Crosses which is explained in that link and couple more things... and also Tensorflow TF.transform which is explained also in the link, but the good part of it is that is all being done in code in the Python Noetbook.

I've learned a lot from this course and I'm still learning more and more. but what I can share here is that the course is really well tought and more important, all the notebooks and code examples work really smooothly, the best online learning experience I've had so far.