I finally got a copy of 'Approaching (Almost) Any Machine Learning Problem'

I finally got a copy of my own book, "Approaching (Almost) Any Machine Learning Problem" and in this live video I will show and talk about some of the chapters and will answer live questions :)

Please subscribe and like the video to help me keep motivated to make awesome videos like this one. :)

Follow me on:

[shekharpandey]

Awesome book by an awesome person for the awesome people!!

[AlexeyMatushevsky]

Received a copy of your book today. I see lot's of code, which is great! Love it. So far so good. Looking forward to read from start till the end.

[bertobertoberto]

Got a kindle copy. I ran through it and will go back through it. I’m not entirely convinced about all the approaches, at least when it comes to tabular machine learning using sklearn. You didn’t use the pipeline abstraction and I really found this essential for enabling reproducible ML in a production environment. Additionally I like the way you split the data into folds. However I didn’t see you following through on that, using it to build out an end to end ML system, mean from data ingestion to deployment. Regardless, it’s a solid effort and I’d be happy if the author or someone can respond to my concerns

[AICake]

I am currently reading your kindle edition. First of all, I need to thank you for writing such an excellent book. I am in the cross validation section and I have couple of questions 1. You are manually shuffling the data before folding. How is it different from keeping shuffle parameter true in KFold/StratifiedKFold method from sklearn ?2. How is your cross validation approach different from sklearn cross_val_score ?

[ravigarg]

Minor Comment: In the discussion of ROC and AUC, you say AUC is a good metric for skewed datasets whereas in reality it is not; one can and should use area under precision-recall curve to be sure on how the model is performing

[idreeskhan]

I wish could buy but due to covid pandemic lost my job cannot offord at the moment bro. Will by later but lots of respect for your efforts.

[nibinjoseph]

Abhishek Is it necessary to know all the algorithms to understand this book?

[namulee]

Sadly, I cannot ship it to Singapore....

[manjunathchittipakala]

when the print book will be available in india

[AlexeyMatushevsky]

Who come up with cover of that book? or if it is - How you come up with that cover? Thanks

[Rahulsircar]

I can't buy it for Rs 1700.Please launch it in India soon.l and share your Patreon id with us.

[mayukh_]

In my opinion, instead of making 1 course, make 2 courses. Because the basic and intermediate things that you talk about, most full time ML engineers would already know and they have to wait till the end for the actual advanced stuffs. While making a separate course for advanced, you can include topics like

1. Transformers
2. CRNN and ocr
3. GAN on NLP and CV
4. Wavenet and Tacotron2
5. Model interpretation for complicated models like CNN and Transformers.
6. Probabilistic Graphical Models
7. Bayesian Networks
8. Associative Rule mining
9. Siamese Networks
10. Zero Shot learning models

Hell I'm even ready to pay 50k for the course if you would include all these.

[gangadharreddy]

How Indian version differ from imported one? How about a kindle version price on india.