I would like to add that the reasons to use a link function is that the values of the predictors don't always have the same range as those in the outcome variable, therefore in order to avoid bias and to truly capture the underlying relationship we use a function as a link in order to map the values from the range of the independent variables to those in the dependent variable. In logistic regression you need the function to have values between 0 and 1, so you use a logistic function to keep that property so that the values of the independent variables can actually match those in the dependent variable.
like half the series went straight over my head but I'm still a bit sad it won't keep going looking forward to rewatching it and understanding everything someday (hopefully) - until then I'll watch anything else you cook up
This brought me back to my masters (statistics) days. Im happy to see the exponential family again. ❤ And hey, thanks for showing some simulation in R. Happy holidays too.
Class. I just got exposed to the exponential family of distributions in my stat theory class this past semester (start of my masters in applied stats) and suddenly a bunch of things clicked regarding GLMs. This vid was a nice recap of that.
So, if p_\theta is of that form for both discrete and continuous variables… should I think of this as, “p_\theta is a probability density with respect to some measure, and when it is describing probability, that’s just because it is with respect to the counting measure”? Or, is that not the right way to think about it?
Don’t get too lost in the sauce, just think of p_theta as representing a parametric distribution with some parameter vector theta. Once the distribution is specified, it can be talked about with the proper terms. No need to get into measure lol
I mean, you don’t have to make things up lol the video is fine without exaggerations. I don’t know what “people” you know that think statistics have always just “existed” but everyone i’ve ever met, correctly; assumes statistics was a human invention, Anyway good video otherwise, fairly informative.
Intro just gave me a heart attack. Crazy rollercoaster but glad you’ll still be around!
You are a true educator, thank you for keeping my love strong for statistics in a time where uni and grad school often crushes your passions
I would like to add that the reasons to use a link function is that the values of the predictors don't always have the same range as those in the outcome variable, therefore in order to avoid bias and to truly capture the underlying relationship we use a function as a link in order to map the values from the range of the independent variables to those in the dependent variable.
In logistic regression you need the function to have values between 0 and 1, so you use a logistic function to keep that property so that the values of the independent variables can actually match those in the dependent variable.
Hey! Don’t stop making videos. You’re doing a very important job! World needs people like you!
Honey, wake up! New very normal video just dropped!
I was genuinely looking for a video explaining GLMs. Thank you!
The equation at 9:42 says e^beta_0, but it should be e^beta_1, a friendly correction! Love the channel, thank you for the videos :)
like half the series went straight over my head but I'm still a bit sad it won't keep going
looking forward to rewatching it and understanding everything someday (hopefully) - until then I'll watch anything else you cook up
This video was so good that I forgot my disdain for statistics. Good job.
This brought me back to my masters (statistics) days.
Im happy to see the exponential family again. ❤
And hey, thanks for showing some simulation in R.
Happy holidays too.
he going to explore guys, not gone forever
lol yeah I was just tired of saying “very normal therapeutics”, I’ll still be explaining stuff y’all
Minor mistake at 06:05 the sign should be a '+' in front of n*log(1-\pi)
Besides that, great video, and a wonderful conclusion to the series!
Class. I just got exposed to the exponential family of distributions in my stat theory class this past semester (start of my masters in applied stats) and suddenly a bunch of things clicked regarding GLMs. This vid was a nice recap of that.
So, if p_\theta is of that form for both discrete and continuous variables…
should I think of this as, “p_\theta is a probability density with respect to some measure, and when it is describing probability, that’s just because it is with respect to the counting measure”? Or, is that not the right way to think about it?
Don’t get too lost in the sauce, just think of p_theta as representing a parametric distribution with some parameter vector theta. Once the distribution is specified, it can be talked about with the proper terms. No need to get into measure lol
Follow up video idea: explain the purpose of the quasi families for GLMs in R
I'm sure this video is very informative, but I though this was about something completely different when I read gay in the thumbnail...
Fantastic!
Yooooo
Christmas came early this year
Great video! But since you only cover the basic concepts it would be nice if you could provide some additional resources.
Sure! What kind of resource did you have in mind?
Merry Christmas. Please do not stop maykingnvideo or delete any video
Well explained, thank you, just subscribed to your channel
Can you for a video on the best books to get acclimated to statistics
I can try to think of a few. It would help me out to know what you want to use statistics for, can you tell me a bit about that?
I WISH I could do that! If i had a 3d printer i’d print soo much stuff!
I think I lack some prerequisites
g(EY) matahamtics
Put a like, now I can watch the video
I mean, you don’t have to make things up lol the video is fine without exaggerations. I don’t know what “people” you know that think statistics have always just “existed” but everyone i’ve ever met, correctly; assumes statistics was a human invention,
Anyway good video otherwise, fairly informative.
I watched this a few hours ago but only now realize that the caption in my inbox says gEY.
i think u need more clickbaity thumbnails :)