Pipelines & Custom Transformers in scikit-learn: The step-by-step guide (with Python code)

Implement custom transformers and pipelines in scikit-learn using python.

[dgory]

Great video! You really understand the nuances of productionisation. A step above other resources on the topic 👍

[imdadood]

This is one hell of a comprehensive tutorial on Pipelines. Thank you for this.

[im_Ashutosh]

Great video Himanshu, I was looking this kind of video on YouTube and I am lucky, I find your channel

[abirkar]

This is brilliantly explained. I was looking at a lot of sources explaining this very topic but none of them did it as lucidly as you. Thanks a ton!

[torque]

very few tutorials on this topic and you explain very well.. I also read you blog on this topic.... thank you so much

[BalaMurugan-cmev]

Nice explanation about the custom Transformers. i have never seen this implementation but you did it. thanks

[CrazyFunCricket]

wow you are rock
the way to explain is tremendous.. please make some more videos.

[cooldudesheks]

Clear, on point and proper do's and dont's that no one would cover! 👍

[gayathrigirishnair]

Thanks a lot for the video that accompanies the wonderful article. I always find articles in general a bit more challenging to grasp. Providing a video to accompany it a great idea. I am now able to follow the article much better.

[swarnachoppella]

Thank you, Sir. It's really helpful

[rakeshbaral]

It was a nice informative video sir. Looking forward for more of the videos like this

[lando]

thanks Himanshu, this tutorial was greatly needed!

[godwinkhalko]

Thank you so much man. You have no idea how badly I needed one of these tutorials.

[aimenbaig]

really loved it, can wait to use this powerful feature <3

[neyagapulasiddartha]

nice video it is very helpful for my course ...it would be great if u make full ML course

[manaschhotray]

Sir, This video was really helpful 🙏🏼. Thank you for all your works.... are you planning to make any more machine learning or scikit learn playlists or videos?

[cooldudesheks]

In case the target is a categorical feature, how would we call CustomTargetTransformer class, since TransformedTargetRegressor would work for Regression problems? Although unlikely, would the TransformedTargetRegressor work for the classification problems?

I am thinking to implement Ordinal Encoding to my target variable in transform() and performing inverse_transform() using the same OrdinalEncoder() class, but I am not sure how to apply that through transformers and Pipelines. Even when I am trying through ColumnTransformer() class, it's throwing an error - ValueError: A given column is not a column of the dataframe - since it's not able to find the "y" when the fit() is called.

Would appreciate inputs on this! 🙏

[brijesh]

@10:17 won't the predict of LinearRegression will be called instead of that estimator class?

[ramendrachaudhary]

1. Why is the fit method left empty? Under what circumstance does fit method need be be defined with some action?
2. X-train and y_train are both passed to the pipeline. Does the transformer apply to both x_train and y_train?

[ramendrachaudhary]

How does transform and inverse trasform work automatically with the customTargetTransformer?? Are the names transform and inverse_transform mandatory in this case?