21. Generalized Linear Models

MIT 18.650 Statistics for Applications, Fall 2016
View the complete course: ocw.mit.edu/18-...
Instructor: Philippe Rigollet
In this lecture, Prof. Rigollet talked about linear model, generalization, and examples of disease occurring rate, prey capture rate, Kyphosis data, etc.
License: Creative Commons BY-NC-SA
More information at ocw.mit.edu/terms
More courses at ocw.mit.edu

[@alizestrelado]

I've learned more in this hour than in a four-months course

[@BonnieX21]

Alize Leal why is this so painfully true
I am sitting here an hour before the online exam due to covid and it’s just mind blowing of how scared I was all semester cause I don’t understand anything but it’s actually the prof 😂😌

[@redabouharrou7950]

i totally agree !!!!!

[@siddharthsvnit]

This is the best explanation of exponential family i have ever come across.

[@stefanogaleano1186]

Never learnt so much in a single lesson

[@richarddow8967]

@29:10 an Army of CDF's that map to 0 to 1 and are invertible. Such simplicity. Love it

[@3ss3ns3]

awsome lecture, the content just flows into the mind without much effort!

[@haoqiu2683]

Best explanation for the exponential family

[@phsamuelwork]

Excellent lecture, horrible recording. Hardly hear anything with max volume and why the camera has to shoot at the prof speaking rather than staying on the slides? It is very distracting and hard to follow.

[@xenosicotte]

I like the way he makes "measures" seem easy to understand

[@Eizengoldt]

What’s a measure

[@bekirbartin2094]

Fantastic lecture. Many thanks for this.

[@michaelamos5576]

I think there's some confusing notations at 27:10, which should be :
The response variable is binary so there is no choic: Y|X is Bernoullli with expected value μ(Y|X) ∈(0,1)
We can't write μ(Y|X) = X^T * β.

[@HerpetologoTropical]

Thank you very much for sharing. 👏

[@AlecPaulssluaPcelA]

Such a good lecture

[@nabilakdim2767]

I think that choosing the distrib on Y is mostly done based on MaxEnt principle, not arbitrarily as said around 19:45.

[@jasonchen3301]

can hardly hear with max volume...

[@siddharthsvnit]

Use ear phones

[@jensharbers5620]

You can also use subs to get the rest, especially if your mother tongue is not English

[@matinhewing1]

Clean your ears...

[@MrSpiritmonger]

I can hear at 30% on my ear phones.

[@harishganesan3575]

i thought my headphone is bust..

[@qiaohuizhou6960]

12:10 very good explanations on what a link function does

[@olivergao7706]

Good course. Just the recording crew needs to realize that in a math class you're not supposed to be staring at the instructor. Instead you look at the board and the slides more

[@kossonouprunelle7576]

Thanks for your presentation. Please in the case we use ordinal logit , should we report pearson correlation and omnibus test value? if it is the case, how to interpret them (for exemple, under or above p value, what it is the meaning). Also shoud we consider the sig level from the table 'Test of model effect' or 'Parameter Estimates' table to say that a relationship between the predictor and outcome variable is significant. I am really looking for ways to interpret, your answer will really help me. thanks.

[@d.langary7000]

1:02:50
Is that really the reason? Why not then define eta(theta,phi) = theta/phi ?
I thought the reason it may not turn into the exponential family may have to do with the fact that c(y,theta) may be not separable.

[@d.langary7000]

1:05:05
It seems he clarifies it later. I just had to be a bit more patient!

[@Volteer]

MIT please provide water and sponges for your professor's blackboard cleaning needs wtf

[@ridwanwase7444]

Poor mit

[@tempvariable]

45:10 I assume h(x) should be 1/sqrt(2pi)

[@TheVSDave]

It is mentioned that "phi is known positive value" (Lecture 23: 16.28 min). I am wondering, if exponential dist. comes under canonical exponential family (or not), where theta = lambda, b(theta) = ln(theta) and phi = -1?

[@not_amanullah]

this is helpful ♥️🤍

[@mantistoboggan537]

Sorry I'm having trouble keeping up, but for the Poisson example, where does that expression for mu(x) come from?
This is terrific by the way- thank you MIT for helping me stay even a little afloat in one of my classes.

[@tempvariable]

Can you tell us which example for Poisson ? The disease example?

[@not_amanullah]

thanks ♥️🤍

[@Random-sm5gi]

5:53 Subtitle should be GLM not GLN

[@xinnywillwin]

I'm in love

[@Hamromerochannel]

i wish GLM is taught this way. Remember how he started, simplying the context of GLM!!??

[@nujranujranujra]

Copy the URL and play it in VLC with volume 200%

[@lasha-georgemath8961]

God damn, You saved my life. Thanks a lot

[@thomasrad6296]

You gotta love the fact that this is MIT and the Volume quality is shite

[@betrayalact2610]

It's just low, I don't think that actually quality is that bad. Also, this was delivered as a live lecture which means sometimes not everything will work perfectly.

[@avasheshkumar5770]

28:55 Seems Accurate hahaha

[@chazzmian]

optimal non-success rate

[@juanfa98]

jefe

[@mirexius]

basically, I got lost after 1:30 xD

[@raymondgooh]

i feel you

[@thalithathomas6512]

🤣

[@freeeagle6074]

Hope you've found your way back after some exploration effort in the forest of GLM.😀

[@OtherMcChicken]

bruv, clean up your blackboard. Your blackboard is whiter than my whiteboard tho

[@chazzmian]

SEXUAL Harassment!

[@jaspreetsingh-nr6gr]

@ 39:07