Hands on Machine Learning - Chapter 7 - Ensemble Learning and Random Forests

An overview of Chapter 7 of the book Hands-on Machine Learning with Scikit-Learn Keras & Tensorflow

Hardware:

[ashmalvayani]

You're doing a great job at what you do. It's really helpful when we can actually code along with you. Let's just hope for more of that type of content. Cheers.

[wilrivera]

Solid videos. I'm following you as I read the book, and it's been incredibly helpful.

[elishabulalu]

thank you so much bro, I have been following the series from the start, It's so cool and lit.

[alexiojunior]

Yeah your tutorials are good and easy to understand i recommend these to those who are new to machine learning and need more of a roadmap well done

[adrinemarieadonis]

I'm gonna follow you from now on. ❤

[pedroabranches]

Man, your work helps me a lot! Thank you!

[gabrielfreire]

Thanks for all the effort and knowledge given on the videos. I do have one questions on min 10:00 of the video you scale the data after it was one-hot-encoded, does it mean that the binary columns get now centred on 0 and variance 1? this would give us values different than 0 and 1, but values ranging from 0 an 1? is that the way to do it?

[sehaj]

Shashank, have you heard about Applied AI course (AAIC) ? Do you have some tips or feedback/opinions..

[pranavgupta]

Isn't bagging/pasting similar to what we do in K-fold cross validation?
Since, both use different subsets of the training data in fitting the model, i.e., k fold cross validation is just another case of bagging/pasting, right?

[fatimak]

bahaha the distraction caused by the porsche 911

[omaralkhasawneh]

why you not complete deep learning part

[fatimak]

commenting for the YoutTube (ML) algo :)

[wilsvenleong]

12:00 Just wanted to point out that your explanation with bagging and pasting is not entirely right.

Bootstrapping means sampling with replacement. It means you take one sample, put it back and draw another sample until the desired sample size is reached. Bootstrap aggregating or bagging means doing this multiple times with mutilple predictors or classifiers.

Whereas for pasting, you sample, don't put it back and sample again until the desired sample size is reached.

The way you explained it was you would sample a certain subset and train on it before replacing it which is not how bootstrapping works. Your explanation would mean that the probability of a sample getting drawn would vary as you will only be replacing the entire subset at the end. In bootstrapping however, the probability of a particular sample being drawn would be equal because after every draw, you will be repacing the drawn sample.


I hope this clarifies things.

[shaelanderchauhan]

Scaling is not necessary for random forests

[pavlostsoukias]

@Shashank Kalanithi I know it sounds stupid what are you drinking?

[orioncloud]

You have a data leakage because of encoding, FYI.

[anamikadassmanna]

It was very poorly done. I was banking on your video and thought of subscribing and getting patreon but this was awefully done... Sorry cant sunscribe also...