Great, your voice is normally distributed

@jesusvelazquezdelatorre8060

3 жыл бұрын

LMAO

This is just great. Well paced stats videos with high resolution, good audio and implementations in code. Can't thank you enough.

That's a very well done video on Box-Cox transformations, I particularly liked the r demonstration and the diagnostic plots. It really bought the message home. Well done!

Woah, this was great. Keep them coming!

Thank you for your great explanation. Looking forward to more videos!

This was really great! Thanks for explaining it at a high concept level. Cleared up alot of confusion for me from the text. Also, the little trivia fact was very interesting! I don't think any of my professors even knew that. Kudos!

Really clear and easy to understand, helped me a lot. And the funny story at the beginning is really interesting :)

I cannot begin to explain how much this helped!

Thank you for this amazing tutorial! You explain something complicated in a way that it's understandable. I hope for more tutorials from you!

Thank you. Your walk through on this topic is excellent.

Nicely done. Thanks for the practical demo!

Thanks for the explanation...this is great!

Great video. Really helped as trying to understand Box-Cox transformation right now

thanks for this amazing video with explanation!

Thanks, this is awesome help for R newcomers! :)

great explanation! thank you!

Just Amazing! Thanks.

We thank so much for saving our assignment with this video

Thank you so much! I understand everything now! You're amazing at explaining things.

@mathetal

5 жыл бұрын

glad it helped 😊

Fantastic. Thank you.

Thanks for your great job!

Thanks!! It helps a lot!

I've seen this video already twice and will many times more.

To the point and very clear explanation

This is excellent--thank you!

Thank you so much for this great tutorial. I have a few questions, could you show us the regression results (coefficients, s.e., t- & p-values) before Box-Cox transformation and what changes happened after the transformation as well as how to interpret the results due to Box-Cox transformation?

Simply phoenomenal

Thank you so much .. You have a got a lovely voice ;)

Just seeing the y^l = X b + e formula written down calmed down a lot of anxieties for me. I still don't understand why using this transformation would be a good idea. Thank you very much for this video!

Thank YOU!!!

Thanks for sharing. Thumbs up.

Thank you so much! :)

Wow! Recommended

This is great. Main takeaway I learned is that if the best lambda is equal to zero then you want to log the dependent variable y. Otherwise, you don't need to log y. In this example, best lambda equals -1.42 so you would use a linear model specification (as opposed to log-linear or log-log). Hope this sounds correct.

thank you!

Very good video! Greetings from Colombia!

clear-cut very good

Nice explanation

Legend

thank you! the textbook i'm reading is very confusing and didn't explain well

Loved the explanation. Can you share the link to the data set used? Thanks!

very nice and clear explanation, can you do robust standard error to fix heteroscedasticity in R?

That's a great video. Thank you. Would you run the model at -1, -1.4242 and -2 and review the R^2 and adj-R^2?

Thank you！a question: what does the 95% line mean?

is there a way to change both dependent and independent variables?

Thank you for a superb video. Could you please share your main code in the comments and possibly make a video for bcnPower where the values are negative or zero as Box-Cox yields an error code. Or anyone share a script which transforms a single variable as noted above?

Didn't realize that the transformation is applied to the dependent response variable (in this case, Butterfat).

I need to transform independent variable as well

I noticed that when changing the lambda range in boxcox, (for example from seq(-2,2) to seq(-3,3)), it actually changes the optimal lambda value. Anyone know why?

Thanks for making this so clear! Great video! I have one question: could you have interchangeably chosen to do a transformation with (Butterfat)^-2 ? What kind of transformation would that be (assuming it's not still inverse transf.)?- sorry if it's a silly question, my stats skills are quite basic

@mathetal

3 жыл бұрын

Box cox transformation is on the outcome variable only and is the more common technique. If you want to transform predictor variables that is also possible and is known as a box tidwell transformation

@saikatkar547

3 жыл бұрын

@@mathetal I have one question ...in a data when we are doing scatter plot between predictor and response and that's showing non linearity ...then what to do ?? Do we transform response variable or predictor variable ?? I mean how to understand wheather we have to transform predictor and response ??

i have a question. If I have a negative response what kind of transformation should I applied for non normally distributes residuals in my linear regression? I already applied scale function in my data.

I haven't tried yet and probably Stack Exchange would be a better place for doing such question but, hope I'm not asking too much... I've read boxcox doc and says it only works with lm and aov entries and raised me couple questions, but simply... So , there is this dataset with x, and y1, y2, y3, yn... is there a procedure to calculate an optimum or, at least, a rounded lambda? Thanks in advance,

Add a prediction part and you would have my 10/10

Thanks Math. It would be more better if you spend few more minutes on summary part(modelling on train and test the model on unseen data and compare the metrics before box cox and after box cox). But still the concept you explained is good

Hi ! :) maybe i missed something but how do you get your normal data after that ? Like the list of all the data ?

Hi, I really enjoyed the video and just had a small question. Is there any particular reason why we should round our value obtained for the optimal lambda? If the optimal value is already available, shouldn't we use the optimal value?

@mathetal

3 жыл бұрын

Usually it wont make much difference either way, but there's no real reason besides simplifying the result

box and cox sound like characters

Hello, thanks for the great video. I am still new to R and I got an error message while doing this test "response variable must be positive". May I ask help how to go about this error? Thanks and keep safe.

First off great video. The problem I have encountered is some of my response variables are negative. This approach wont work with those pesky negatives. Any suggestions?

@MintMcCloud

4 жыл бұрын

Hi, I see you commented this 9 months ago, any chance you figured it out? I'm currently having the same problem :(

Can it be used as for arc sine transformation???

Hi, why is the value range for Residuals vs Fitted without the transformation from 3.5 to 5.5 and with the transformation from 0.18 to 0.28?

The notations are a bit confusing: how may y to the power lambda on the left hand side could be equal to log(y)? This is because the y on the left hand side should write y^{(lambda)} instead of y^lambda.

Hello, can someone explain why Butterfat^-1 a.k.a. y^-1 is performed, rather than (y^-1 - 1)/-1 so y^lambda-1/lambda as in the formula from the beginning. Thank you

I'm probably just really confused, but once I get my lambda value, how do I actually transform my data set. I understand visualizing those plots and what not, but does it spit out a new table of transformed values somewhere?

with the transformed data can i diagnostic through Shapiro's and Lavene's tests or only through the plots?

@mathetal

4 жыл бұрын

yes you can still check through statistical tests like Shapiro's and Levene's

I can't find an inverse-Box-Cox function that v3.6.2 of R will load for me. Can somebody help?

i love you

please can you exaplain me in the histogram in box cox trans

Homoscedasti.... vwey... what??? I never know there are assumptions to Linear regression..

Jesus. It’s like reading a Wikipedia stats page. No idea what’s going on here

Voice is too hot, got completely distracted by it lol