Power Transformer | Box - Cox Transform | Yeo - Johnson Transform 

Sir, you are much better than Krish Nayek whatever I seen....

This is the first video that I've watched on your channel. You explain things so patiently and clearly like no one else. Liked and subscribed. Thank you so much.

Great ..Teaching...surely your channel will cross ...1 million mark in future and it will happen suddenly..keep going bro.

you are true legend...best teacher i have seen ever..

This channel deserves at least 1-2 million subscribers !! I have subscribed :)

Best Teacher ever!!!

Underrated content for sure

i am big fan of your teaching sir. saare ke saare concepts aise yaad ho rhe h ki chahe na bhul paau so thank you so much again for creating such a great content.

Day by day the course is being interesting and little difficult also😅

Bahut Mast Video hai... Series bnaa Do.. Mjaa aa jayega Sir🔥❤️❤️

I only like two ways of starting a video. One is yours "hello guys welcome to my RU-vid channel" and the other is shoaib akhtar's " welcome to shoaithar yutuchanel"

Sir, you deserve millions of likes for each and every topic

Its awesome video, my understading around this concept improved a lot.

Love your videos

Great way of explaining. Thanks for nice video

Very good teaching sir, thanks😀

Great video on power transformation. Really appreciable.

Really Valueable .

why are we using plt.subplot(121) and 122. what is the meaning of 121 and 122 here ?

Great

crystal clear!! thank u so much...

Thank You.

sir! salute to you

Thank you so much sir

Fantastic video

In Yeo Johnson transform formula there should be log(Xi +1) instead of log(Xi) + 1 . For Xi >=0 . Is it correct? Correct me ,if I am wrong.

Great 😃👍

Just waqnted to let you know thatI am following your complete playlist of 100 days of ML. The videos have benefitted me quite a lot. Just had one small doubt- In your previous videos of the playlist, you always recommend to do train_test_split before doing any scaling. Specifically, you taught us to fit the object of scaling on X_train, for example, sc.fit(X_train). Then, we applying the parameters learnt from fitting onto both X_train and X_test by using transform feature. My doubt is- In this video, while practicing the code side by side along with the video, I noticed that the Lambda values of X_train and X_test are different. What should we do in the case when the value of a parameter after fitting on X_train differs quite a lot from X_test fitted parameter values? Please suggest (Not only is cases of transformation but to only scaling also) as you are an experienced subject expert of ML.

Sir, please confirm which one should be in sequence (filling missing values) -> (Standardization scaling i.e -1 to 1 / Normalization scaling i.e 0 to 1) -> (Power/Function Transformers ) OR (filling missing values) ->(Power/Function Transformers ) -> (Standard/Normalization scaling) OR is it necessary to do both scaling and transformations.

sir, agr mere data me outlier hai to power transformation ke pahle mujhe us outlier ko change kar na chahiye ki nahi

bahut khub ....thank you

finished watching and coding

Thank you sir

what if the initial data has both numerical and categorical columns

thank you so much

is this transformation used in the industry? I havent seen anyone using it so far..

14:55 ‼I dont thik they are ordinal exponential powers, lembda are the parametersr to put in box-cox function.Plz confirm?

Sir, how to decide value of lambda?

Sir pls reply if you read this comment - if we have bimodal or multimodal distribution of a feature then is it necessary to make it unimodal to improve necessary.. if yes then pls name the method through which we can do it

You should have scaled the features before running LR. As LR gets impacted by different scales of features

Nice content ! And thanks for ur efforts. Do you have the code file in comment section?

sir , how does box cox know the best approximation of normal distribution? is it using np.skew or some other formula?

so we should apply scaling before transformation ?

Sir, scaling kerne baad transformation apply kerte hain or pehle ?

one thing i want to ask Box cox is applied on Y (response) not on X so how here its applied . Please clarify it once

best

Sir agar target value skewed distribution me h to kya usme log transform laga kar usko normal bana ke use kar sakte h kya?.. Please sir reply karna🙏🙏🙏

in some cases like linear regression or logistic regression or svm, we need to scale y(target var) also. Similarly, in those cases, can we apply transformations (like yeo-jhonson) also if it is not a normal distribution ?

Can i ask you why half of your speech is in Englesh and another half is in Hindi?

Sir why you use 'r2_score' instead of 'accuracy' function from 'sklearn.metrics'

thank you sir