Multiple linear regression in R

Multiple linear regression is important and complicated. Let's get started! Note: I misspeak at about 9:27, stating the null hypothesis incorrectly. It should be that there is no linear relationship between the variables.
#regression

[@EquitableEquations]

You can find material supporting this vid (and others) at github.com/equitable-equations/youtube.

[@Nothingimportant1]

It was one of the best videos on youtube. That last interpretation is perfect.

[@EquitableEquations]

Thank you!

[@martinalcala5914]

Thank you very much for your videos. I've watched the ones regarding the multiple linear regressions and they've solved lots of doubts that I had!

[@Nothingimportant1]

I have my stats exam tomorrow and plan to learn R fundamentally by watching your videos. You do a great job.

[@neiltalbert7091]

Thanks for the info on Cook's D. It's probably useful when sample sizes are relatively small and outliers could be disproportionately influential.

[@valaiopwep7990]

Hey thanks for the vids they're a lifesaver, im taking econometrics this semester and as someone with not much experience in stats or programming this helps a lot

[@AlanCheun]

you make good videos, i'd just pad the ends of your video with a few seconds just to let us absorb the final statement and anything on the screen that just changed, youtube will prompt an ad or jump to another video before I can pause or see what just changed

[@calmseeker5501]

helpful videos of all!

[@Victoriataieb]

I wish you can make statistics using R series , I found your explanation straightforward and easy to understand , thank you so much

[@EquitableEquations]

Here's my intro stats playlist. It makes heavy use of R
ua-cam.com/play/PLKBUk9FL4nBalLCSWT6zQyw19EmIVInT6.html

[@Mahdi-Abd-Allah]

Thanks you so much for explaining

[@romanvasiura6705]

Thank you!
I guess we should correct a little bit this code (rows 18-19)
new_data

[@EquitableEquations]

You’re right! It’s fixed on GitHub.

[@haraldurkarlsson1147]

I know there are multiple ways of writing the multiple regression line but I am still a bit confused. Since you are using the approximation sign then is the y not y hat? Also does not each x_i have its own error that is epsilon_i. Therefore, in your expression should the epsilon subscript not be k?

[@buffaloperformanceandanaly1431]

Would there be any difference in the code if there is a positive relationship between one variable and a negative relationship with the second variable? I'm looking at athlete exertion, sleep, and fatigue. Thanks!

[@EquitableEquations]

Hi! There's no difference at all if one relationship is negative and the other positive.

[@saminrasi]

Hi,
I really like the points you mention in your lectures, very helpful!
a couple of questions
1. looking at ggpairs graph you mentioned y values doesn't seem to be dependent on the value of x1 or x2 , how did you come to this conclusion?
2.have you also covered nonlinear regression in this channel?
Thank you!

[@EquitableEquations]

HI! The main points of the ggpairs plots are (a) the relationships y~x1 and y~x2 are linear and (b) the *spread* of the y-values doesn't seem to depend on the x-values. In the latter case, I'm seeing that the y-values in each of the bottom two plots are roughly equally dispersed for each value of x.

[@EquitableEquations]

I'm building a series of vids on multiple regression right now. I expect to cover polynomial models in the next few weeks and generalized linear models after the new year.

[@saminrasi]

@@EquitableEquations 1.Thank you , now understood what you were talking about!
2. great! will subscribe :)

[@mardzj]

at 9:27 shouldn't the null hypothesis be that there is no MLR?

[@EquitableEquations]

Yes you're correct. I misspoke at that point.

[@yaweli2968]

Is there a site where you have all your loaded video lectures in R?

[@EquitableEquations]

Thanks for asking! You can see the whole collection here:
ua-cam.com/play/PLKBUk9FL4nBYpUKszG4edyAiM9aeTT1yv.html