/sci/ - Science & Math » Thread #12492260

6KiB, 364x469, chi.gif

View Same Google iqdb SauceNAO

Self-teaching stats

Anonymous Mon 21 Dec 17:19:04 2020 No.12492260 View Reply Original Report

Quoted By: >>12494874 >>12494911 >>12495211 >>12498027 >>12498175 >>12501694 >>12504459

I looked on the /sci/ guide and there doesn't seem to be a robust stats sections. I want to learn statistics on my own because my uni Intro to Stats class wasn't that great.

What are the best resources for a bit of autodidacticism?

Anonymous

Anonymous Tue 22 Dec 2020 03:37:49 No.12494284 Report

Quoted By:

bump

Anonymous

Anonymous Tue 22 Dec 2020 03:55:46 No.12494325 Report

Quoted By:

I'm interested as well. What's a nice statistics book that doesn't end in absolute beginner retard-tier, and which teaches modelling without becoming schizophrenic math autism full of useless proofs for "elegance" "beauty" or such nonsense?

Anonymous

Anonymous Tue 22 Dec 2020 08:38:53 No.12494859 Report

Quoted By:

stats is not very rigorous
for example, the monty hall problem

Anonymous

Anonymous Tue 22 Dec 2020 08:48:35 No.12494874 Report

Quoted By:

>>12492260
there was that book of intro to stats in R that was not bad...

Anonymous

Anonymous Tue 22 Dec 2020 09:03:16 No.12494911 Report

Quoted By: >>12497914 >>12498190

>>12492260
You need to learn probability first.
There's a very good series of lectures from Harvard:
https://youtube.com/playlist?list=PL2SOU6wwxB0uwwH80KTQ6ht66KWxbzTIo

Anonymous

Anonymous Tue 22 Dec 2020 11:57:55 No.12495211 Report

Quoted By: >>12495520 >>12498030

>>12492260
>X^2 test
>only stat test i dont give a shit about

Anonymous

Anonymous Tue 22 Dec 2020 14:46:37 No.12495520 Report

Quoted By:

>>12495211
Why, they're easy and quite useful when dealing with nominal subjects. We use them a lot in biology and psych

Anonymous

Anonymous Wed 23 Dec 2020 02:58:48 No.12497914 Report

Quoted By:

>>12494911
Thanks, I'll look into this series

Anonymous

Anonymous Wed 23 Dec 2020 03:02:54 No.12497919 Report

Quoted By: >>12501436 >>12505132

statistics by witte and witte seems good

Anonymous

Anonymous Wed 23 Dec 2020 03:43:22 No.12498027 Report

Quoted By: >>12499278

>>12492260
I took a biostatistics course last semester that used "Biological Statistics for the Biological and Health Sciences" by Triola. The e-book with the included video presentations were comfy and top tier. It also includes access to their testing software that provides you with as much practice as you want.

Anonymous

Anonymous Wed 23 Dec 2020 03:44:29 No.12498030 Report

Quoted By:

>>12495211
Chi-squared tests are some of the most important in research dealing with human subjects

Anonymous

Anonymous Wed 23 Dec 2020 05:04:23 No.12498175 Report

Quoted By: >>12499278 >>12501476 >>12504459

>>12492260
As others mentioned, probability theory is kinda forefront. I would say "All of Statistics" is a bible-book that requires some math background. Elements of Statistical Learning is a good intro text past that. Start there. But following:
>Statistical rethinking
fucking this will do a number on you. Especially the golem of Prague, etc. Great way that teaches you to learn how to think about statistics
>doing bayesian data analysis
fucking phenomenal. I can recommend enough after learning a good stats basis.
>A mathematical theory of communication
>Probability Theory: The Logic of Science
Read the Shannon paper, then read the most underrated motherfucker in stats and logic and information, ET Jaynes. Get your goddamn mind blown by entropy and information related to probability. This will not help your stats, just enrich you to reality.

Anonymous

View Same Google iqdb SauceNAO 1607972046921.png, 113KiB, 686x526

Anonymous Wed 23 Dec 2020 05:10:40 No.12498190 Report

Quoted By: >>12499316

>>12494911
not OP, but thanks for the link. just watched the first of these. what's the probability i make it to the end?

Anonymous

Anonymous Wed 23 Dec 2020 15:32:33 No.12499278 Report

Quoted By:

>>12498027
>>12498175
Thank you so much. So far I've just learned about null hypothesis testing, t/z tests, ANOVAs, and chi-squared tests. I don't feel as though any of it was explained well and the professor just wanted us to be able to compute data sets (same with every math class I've ever taken, computation dominates all else and you never learn the reason why you even do this math in the first place). These suggestions should help a lot.

Anonymous

Anonymous Wed 23 Dec 2020 15:51:24 No.12499316 Report

Quoted By: >>12499588

>>12498190
0.01

Anonymous

Anonymous Wed 23 Dec 2020 17:47:17 No.12499588 Report

Quoted By:

>>12499316
p < .01

Anonymous

Anonymous Thu 24 Dec 2020 03:21:12 No.12501436 Report

Quoted By:

>>12497919
I missed this recc, thanks. I looked at the table of contents and seems to cover a lot.

Anonymous

Anonymous Thu 24 Dec 2020 03:45:13 No.12501476 Report

Quoted By:

>>12498175
>Statistical rethinking
>Probability Theory: The Logic of Science
Thank you anon so much.

Anonymous

Anonymous Thu 24 Dec 2020 05:30:28 No.12501694 Report

Quoted By: >>12505074 >>12505179

>>12492260
Look up professors of research stats:
http://bactra.org/notebooks/teaching-statistics.html

He has his own books like modern regression http://www.stat.cmu.edu/~cshalizi/TALR/ and this: http://www.stat.cmu.edu/~cshalizi/ADAfaEPoV/

If you want to learn a bit of probability try this: http://www.cs.cmu.edu/~odonnell/papers/probability-and-computing-lecture-notes.pdf or go on MIT's OCW and take their probability courses they are all open.

You don't need to know much about probability to do stats, probability tells you how likely something is to occur whereas stats is reverse engineering probability given an outcome what is likely to have generated this outcome?

Anonymous

Anonymous Thu 24 Dec 2020 07:31:10 No.12501952 Report

Quoted By:

i think this looks promising
https://drive.google.com/file/d/1VmkAAGOYCTORq1wxSQqy255qLJjTNvBI/view
(found on https://projects.iq.harvard.edu/stat110/home)
but for some reason the authors didn't make the pdf downloadable reeeeee

Anonymous

Anonymous Thu 24 Dec 2020 15:51:59 No.12503089 Report

Quoted By: >>12503096 >>12506831

I NEED A CHART PLEASE.

Anonymous

Anonymous Thu 24 Dec 2020 15:53:15 No.12503096 Report

Quoted By:

>>12503089
Some anon needs to do this and put it on the wiki
Will save us brainlets

Anonymous

View Same Google iqdb SauceNAO bed .jpg, 84KiB, 642x900

Anonymous Thu 24 Dec 2020 18:42:34 No.12503572 Report

Quoted By:

Ich bin eben aufgewacht. Ja moin. Frohe Feierlichkeiten, meine Mädels.

Anonymous

Anonymous Thu 24 Dec 2020 23:27:47 No.12504459 Report

Quoted By: >>12505067

>>12498175
based as fuck (though I haven't read the Jaynes). The Mclreath book completely changed how I looked at stats. I'd also add McKay's text on information theory, machine learning, and bayesian stuff to the list.

>>12492260
OP , it really depends on what you want stats for and what your background is. If you don't tell us, it's impossible to make a recommendation.

Do you want to learn mathematical stats to construct your own estimators? Are you a scientist looking to understand the common tests (ANCOVA/ANOVA, t-test, etc.)? Do you want to go into machine learning?

Anonymous

Anonymous Fri 25 Dec 2020 02:50:32 No.12505067 Report

Quoted By: >>12505074

>>12504459
>what your background is
Undergrad Bio/Psych major

>depends on what you want stats
I've always felt betrayed by how the education system teaches math, so I just wanted to understand it deeper than learning how to merely compute math problems and data sets.

I guess I also just want to develop a robust toolset so I'm versatile. I have an annoying amount of interest which makes it difficult to describe what I want the stats knowledge for, specifically. To some extent I wanted to get into UX research, but there are plenty of other fields too; too many to name.

Anonymous

Anonymous Fri 25 Dec 2020 02:53:20 No.12505074 Report

Quoted By: >>12505179

>>12505067
UX research is free from CMU (google)
If you feel betrayed then watch Wildberger lectures about math
If you want to learn statistics, literally see this guy's notebooks http://bactra.org/notebooks he is a research statistician with a physics PhD, so he approaches the subject from applied math meaning all the bullshit is removed. >>12501694

Anonymous

Anonymous Fri 25 Dec 2020 03:27:46 No.12505132 Report

Quoted By:

>>12497919
Witte is absolute trash. It treats you like a retard and doesnt even show how the formulas are derived.

Anonymous

Anonymous Fri 25 Dec 2020 03:52:39 No.12505179 Report

Quoted By: >>12505201

>>12505074
>>12501694
Thank you. I keep forgetting about MIT's OCWs. That Bactra site seems quite useful, especially the recommended reading lists. I believe the Wildberger lecture page will be a great boon as well.

Anonymous

Anonymous Fri 25 Dec 2020 04:06:17 No.12505201 Report

Quoted By:

>>12505179
The bactra guy is here: http://www.stat.cmu.edu/~cshalizi/
Wildberger actually has 2 youtube channels there's 'wildegg' and his 'insight to mathematics' and there's a stats/prob introduction there too

However if you do his total meme rational trig, you will realize this is also related to statistics and you write your own rational statistics. Don't think his wacky rational math 'won't help you make money' or any kind of other bullshit, it helps you understand the very difficult measure theory/real line measurements kind of stats but from a simplified abstraction using rational analysis (his algebraic calculus course). Anyway that's my rant enjoy I do graduate stats all day for a living in a medical lab and I don't have a degree.

Anonymous

Anonymous Fri 25 Dec 2020 16:48:21 No.12506831 Report

Quoted By:

>>12503089
this

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches