[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 47 KB, 580x584, regression.png [View same] [iqdb] [saucenao] [google]
9373282 No.9373282 [Reply] [Original]

Can anyone recommend any books on statistical analysis? Specifically regressions if possible.

>> No.9373287

>>9373282
no

>> No.9373291

>>9373282
Introduction to Econometrics by Lahiri

You'll become a fucking god of regression if you read the whole thing

>> No.9373594

>>9373291
Great, I'll check it out.

>>9373287
Thank you

>> No.9373611

>>9373594
Note that it is not the easiest read - very dry, very low on plain language explanations

>> No.9373815

theres never any love for regression on /sci/

>> No.9373876
File: 3.75 MB, 420x420, Avian Rain Dance.gif [View same] [iqdb] [saucenao] [google]
9373876

>>9373291
Who knew a book on regression can be so forward thinking?

>> No.9374389
File: 92 KB, 1400x807, Screenshot from 2017-12-17 09-46-17.png [View same] [iqdb] [saucenao] [google]
9374389

I was working through some of "Statistical Rethinking: A Bayesian Course with Examples in R and Stan" a while ago. It's a good book with a focus on intuition. No need to use R or the author's libraries if you already have some familiarity with Stan; I was doing the examples in Mathematica just fine.

>> No.9374417

>>9373815
maybe because it's mathematics' equivalent to social science

why the fuck would there be any love for something that amounts to "lmao i GUESS a linear separation in some n-dimensional space categorises my data properly XD LMAO bro XD"

>> No.9374425

>>9374417
>Regression is always linear
Confirmed know-nothing.

>> No.9374428

>>9374425
you're right, in my rush to badmouth statisticians I made a mistake. I shouldn't have implied that all regression is linear

my point still stands that regression of any kind is unfounded biased bullshit and that statisticians should kts

>> No.9374437
File: 47 KB, 826x264, Screenshot from 2017-12-17 10-19-55.png [View same] [iqdb] [saucenao] [google]
9374437

>>9374428
My dude you might as well say ALL statistical models are bullshit because regression is just a special case of the more general technique of Bayesian data analysis. Are there oversimplified models? Yes. Do they still serve their purpose in a myriad of industries like insurance and finance? Yes. Saying regression is bullshit is just about the biggest giveaway that you are a freshmen who barely knows the first thing about statistics.

"All models are wrong, but some are useful."

>> No.9374440

>>9374437
>Saying regression is bullshit is just about the biggest giveaway that you are a freshmen who barely knows the first thing about statistics.
no, I'm a grad student who had exactly two (2) statistics courses in his life before he realised he had no interest in being a glorified witch doctor making predictions based on unfoundable shaky grounds

>My dude you might as well say ALL statistical models are bullshit
which is the case

statisticians get OUT

>> No.9374453

>>9374440
I don't know what to tell you except that you're dumb. Every scientific field uses stats to analyze experimental results.

Yeah there's a fair bit of witch-doctoring and domain knowledge that goes into crafting a particular model as far as hyperparameter selection goes but even things like that get addressed in Bayesian nonparametric modelling. You probably just got off with the subject on the wrong foot with faggy frequentist shite which yeah sucks ass. Try some Bayesian stats and see if you still think it's all a bunch of BS.

>> No.9376287

>>9374440
>>9374428
>my point still stands that regression of any kind is unfounded biased bullshit and that statisticians should kts
>I'm a grad student who had exactly two (2) statistics courses in his life

stop embarrassing yourself you fucking brainlet

>> No.9376685

>>9373282
good choice anon, once you get the basics of linear regression i would highly recommend to move straight to Applied Logistic Regression by David W. Hosmer

>> No.9376709

>>9374453
>frequentism means that it is bad

popsci faggot detected

>> No.9376729

>>9374428
>virgin mathematician
versus the Chad Statistician. It's okay, you'll go back to college just to get that sweet sweet stat degree later.

>> No.9377178

>>9374440
This is one of the most pathetic posts I've ever seen on /sci/

>> No.9378003
File: 26 KB, 300x300, alpha-as-fuck_fb_3188881.jpg [View same] [iqdb] [saucenao] [google]
9378003

>>9376729
>tfw getting a masters in statistics
>tfw when you are currently ascending Mt. Olympus

>tfw Chad statistician in the near future