[ 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: 106 KB, 1334x750, IMG_0217.png [View same] [iqdb] [saucenao] [google]
9175738 No.9175738 [Reply] [Original]

What's /sci/'s opinion on statistics?

Regards, guy who wants to do a statistics PhD.

>> No.9175783

I think it's a great application of probability theory and it's really one of the most applicable things you can learn - statistics is everywhere.

Don't know much about PhDs in Stats but I checked the placements at a few of the top schools and the placements are really great

>> No.9175789

God tier, it's applied mathematics but stats majors actually get taught proofs and rigour.

>> No.9175949

>>9175738
I'm a PHD student in statistics. It's a decent branch of math since all the questions have an appearance of being applicable but people still want you to rigorously prove them. However depending on your personality, teaching could be a major pain in the ass. You think undergrads taking calculus are bad? Undergrad in statistics courses are worse. I've literally asked a student about the binomial distribution and she looked at me like I was chanting voodoo. In calculus you can shut a student up by writing down the right equations, but undergrad stat students are unironically scared of equations and want you to explain in words.

>> No.9175963

What are the best books to learn statistic?
I want to have a sold knowledge of the field befofe applying.
Bonus point for books with solutions

>> No.9175968

>>9175963
It depends on what part of statistics. Are you looking to learn on basic statistical inference? Stochastic processes? Regression? Finite measure theory? There's different textbooks on all of them.

>> No.9175995

>>9175963
I recommend Casella and Berger:

https://www.amazon.com/Statistical-Inference-George-Casella/dp/0534243126

There is a solution manual with most of the problems solved which is separate from the actual book. However it is also extremely easy to find a pdf of the solution manual through google.

>> No.9176025

>>9175738
Great on its own, even better if you go into actuarial science and can deal with the bullshit that is businessfags.

>> No.9176422

>>9175738
Amazing subject, incredible applications, academia is full of autists

>> No.9177366

>>9175738
I have worked with some people with higher education with statistics and a few were total fucking morons.

Can someone explain to me how you can get a master's or higher in a difficult subject but still be completely inept. Is this common in other higher level professions? How do these people not get weeded out?

>> No.9177372

stats to math what engineering is to physics? I feel like if you're really good at it - phd caliber with interesting research - the job opportunities are probably phenomenal

>> No.9177386

>PHD in statistics
??? Why would you go to graduate school to study mean median and mode

>> No.9177388

>>9177386
Some people are sad.

>> No.9177412

tbqh, it is way fucking harder than deterministic math.

You always need to take into account the correlations between different variables in any given system of equations, ans well as their different distributions and all the implications that come with them.

Maybe I'm just a brainlet but econometrics is one of the hardest classes i'm taking right now

>> No.9177440

>>9177366
I'm not sure in what area those people you're talking about are deficient in. However I can say that most graduate statistics courses don't require the level of rigor most graduate math courses do. You can feasibly pass the required curriculum without always thinking like a proper mathematician.

>>9177386
I could list out a bunch of materials to show how much of the subject you've neglected to mention, but instead I'm just going to ask you a simple question:

For an iid sample obtained from a normal distribution with unknown parameters, which of those three estimators should I use to estimate the true mean and why?

>> No.9177450

>>9177440
>which of those three estimators should I use to estimate the true mean and why?
If you need to ask me a simple question about mean median and mode then in all Likelihood statistics isn't for you.

>> No.9177452

>>9177450
I know what the answer is and why, but what about you? I'm asking if you can answer it.

>> No.9177456

>>9177440
it was a joke (and a pretty decent one at that), dude, chill

>> No.9177464

Statistics was the only math class I ever struggled in. Not sure why. I just couldn't wrap my head around some of those concepts and I kept making small mistakes on exams.

>> No.9177478

>>9177440
I didn't mean particularly deficient in the subject itself (I don't know enough to prove or disprove their work). They just struggled with very, very basic parts of the job (i.e. "How do I pull in the data?"). Stuff that any normal person would figure out within a few months. Stuff they used daily. Yet, these people were in charge of work where mistakes would be very expensive. I just find it really hard to believe they were doing a good job. Management couldn't call bullshit either because they couldn't disprove any of it.

>> No.9177554

>>9177478
They couldn't pull in data? Was it in a specific format or something that prevented them from googling the answer? Any decent department would give assignments requiring their students to read txt files into R.

>> No.9177557

Is this the field we use to prove minorities and women are inferior to the white man?

>> No.9177559

Took a baby upper division statistics course for Elecreical engineers; was definitely hard since it's so different from what I'm used to. The course was very interesting though, and statistics has a lot of real world applications. I can't say if it's harder than math but it's definitely tough.

>> No.9177560

>>9175963
Casella and Berger is your bible. But if you wanna get good at the programming side check out A Course in Statistics with R.

Also this autistic math professor at a podunk university made this badass site with the best explanations of anything statistics-related I've ever seen.
randomservices.org
Check out the apps (!)
>>9177412
Yeah, I actually want to make some cool econometrics for my PhD. I'm really into urban development. But my uni has no graduate economics department. Just Biostatistics. I think I'm gonna move.

>> No.9177562

>>9177559

>can't say if it's harder than math

t. Kate Perry

>> No.9177563

>>9175738
the most red pilled subject

>> No.9177567

>>9177563
Like half of the PhD student body are Indians and Chinese at where I am, and program's ranked within the top 20. The proportion among Master's students is even higher.

>> No.9177572

>>9177567
same here

>> No.9177576
File: 75 KB, 1024x962, tfw cant speak le fraçoise.jpg [View same] [iqdb] [saucenao] [google]
9177576

>>9177567

>half of the PhD student body are Indians and Chinese

https://www.youtube.com/watch?v=F6BUZZ3qvZM

>> No.9177581

I want to go to Duke for my Stat PhD.

But if I have an MS Math and then finish out an MS Biostatistics are they gonna consider me "too prepared"?

I heard they don't like to recruit guys who've been sitting around academia super long.

>> No.9177585

>>9177581
why would you get an ms, get another ms, then get a phd?

>> No.9177651

>>9177581
Duke is primarily Bayesian, be warned

>> No.9177656

>>9177651
What's your issue with Bayesian methods?

>> No.9177657

>>9177656
nothing, just letting him know it's not your average Stats program

>> No.9177671

>>9177585
Because I got depressed and failed a class in Spring so won't get accepted for Fall '18.

I got really bummed out studying pure math and not getting to study any of the statistics I wanted. Sounds dumb but it happens. I kept all the proofs on my desk and didn't send them in cause I was depressed studying pure abstraction instead of cool statistics stuff.

>> No.9177676

>>9177671
studying pure math sucks unless you're...ya know. do you already have the one ms? go get a job dude school fucking blows.

>> No.9177686

>>9177676
Yeah, I'm just surveying my options.

I always wanted to learn stats but my school only offers pure theoretical math, and then Biostats.

Fun fact: the department chair actually told me we're allowed to take 12 hours of Biostats and count it towards the MS Math degree but it wasn't listed anywhere on the department site, no advisor told anyone that, and when I told my cohorts none of them even believed me until we asked the secretary.
And he only told me that after I'd signed up for courses for this Semester. Fucking douche, I wish I could have taken Survival Analysis or something cool instead of some shitty Numerical Linear Algebra class when I've already taken 2 other Linear Algebra classes. Survival Analysis is an awesome topic too. Used all the time in actuarial science.

>> No.9177692

>>9175738
Just do probability in pure math.

>> No.9177694

>>9177686
>survival analysis
if you're getting an ms in math (i am too atm) can't you just pick up something like that easily? i think math is a better degree in preparation for a stats phd than stats would be. you'd have to study a little harder to catch up on the basics, but i think after that your ability to dive deep would outmatch someone whose preparation was bs stats -> phd stats

>> No.9177717

>>9177554
I mean it in the context that one person couldn't build a basic query. Straight from the database. For comparison, it would be like the senior mechanic at a shop asking how to change a tire. Building it from a text file would be much harder in this instance.

>> No.9177876

>>9175738
this year on college i have 5 statistics subjects
any statistics books anyone can recomend?

>> No.9178196 [DELETED] 

>>9177388
>>9177386

Reported

>> No.9178206

>>9175738
porn isn't allowed on this board

>> No.9178219

I find math interesting but I didn't enjoy statistics. I can't put my finger on it but there's something about it that isn't as intuitive.

>> No.9179621

>>9175949
Statistics is one of the most unintuitive fields ever. Even after proving something and having it explained, it's still hard to believe/understand it. I don't blame them!

>> No.9179642

>>9177563
what does this mean

>> No.9179645

>>9175738
No one should be able to graduate high school without a statistics course. It's one of the most important things anyone could learn to function in today's society.

>> No.9179647

>>9175738
I just got accepted to the undergraduate stats program at UWashington Seattle

any tips?

>> No.9179649

I didn't know statistics was so esteemed by /sci/. It has to be one of the most boring, dry subjects

>> No.9179697

>>9179649
it's boring yeah, but it also lets you categorize and manage uncertainty and work with unpure numbers, something that's super useful to any pure mathematician when they leave their house

>> No.9179748

>>9175738
Applied statistics is a great subject.
The theory behind statistics is fucking cancer and should be avoided at all costs.

>> No.9179757

>>9179748
Why?

>> No.9179782

>>9179649
Because it's useful no matter your field to sift through all the data you get.
Terrible analogy time
>Math=Letters
>Stats=Grammar
>Science=Essay

>> No.9179785

>>9179649
I thought this way but later I realised that statistics is actually a very deep subject.

As a field within pure math perhaps not so interesting; it is more like physics - conceptual issues are more prominent.

These might get you a sense of the possibilities

ET Jayne probability theory the logic of science

Gelman Bayesian Data Analysis

>> No.9180550

>>9179647
My tip to you is to double major in math; to be a real statistician you kind of need a bachelor's in math.
My other tip is to stay on for that 5th-year master's in stats. Seriously. Washington is probably the best statistics program in the country.

>> No.9180556

>>9179748
The theory behind statistics is fucking awesome you little shit.

>> No.9181222

>>9180550
I didn't know there was a 5th-year master's in stats. I'm double majoring in Computer Science, but minoring in Math. Thankfully, the stats program requires Real Analysis (from the math department), so hopefully I'm not missing out on too much.

>> No.9181234

Have statisticians a future when you take automisation and machines into account?

>> No.9181241

>>9181234
IMO statisticians such as actuaries might not have much of a future because of automisation, but statisticians in computing / data science / machine learning definitely do have a future.

And I'm actually not too sure about the actuary aspect either, not completely sure if that can be automated (I would think if it could be, it'd have been automated by now)

>> No.9181256

>>9179748

t. sociologist

>> No.9181363

waste of time. just learn deep learning.

>> No.9182849
File: 50 KB, 640x592, 1505089753976.jpg [View same] [iqdb] [saucenao] [google]
9182849

"Statistics are like the lamp post to the drunkard; used more for support, than illumination."

>> No.9182881

My country is badly lacking in statisticians so they made 300k starting meme true but at BSc (no PhD required!) level if you majored with a focus on statistics.

>> No.9182882

>>9177386
>statistics PhD
Some people are unstoppably autistic

>> No.9182953

>>9175738
ACTUAL statistics done with rigour is really fucking difficult for me. I'm really fucking struggling do learn it.

>> No.9182982

>>9177452
It was an estimation joke, anon. Likelihood as in maximum likelihood.


Undergrads, i swear..

>> No.9183734

>>9182881
Which country?

>> No.9183748

>>9177464
Is taught by a brainlet teacher, stats can be tough.

Also, dry stats can be hard - lubed up with biology, it's not bad. (I've taken a few different stats courses - for fun) Learning ab applications is always easier for me. Why is this info useful, etc.

>> No.9183755

>>9183734
Srsly

>> No.9183756

>>9179649
>boring
?????¿¿¿¿¿

>dumbfounded

How?? It's so ..... applicable

>> No.9184675

>>9175738
What is the best stats-related job out there?

>> No.9184688

>tfw the big-dicked alpha machine learning PhDs get paid 3 times more than you and can predict things much more accurately than your beta little hand-crafted generalised linear models

>> No.9184692

>>9184688
Linear model? More like lambda, sheesh

>> No.9184712

>>9175738
>study to draw conclusions of things everyone already knows except you have a degree now
>still cant say there is a disparity between intelligence and races or anything having to do with race because that would be 'racist'

Fuck you faggots

>> No.9184714

>>9184712
There are better questions to answer other than white male inferiority complexes
>neurotypical spreg
>aha

>> No.9185009

>>9184688
what about bioinformatics, ms?
virigin or chad?

>> No.9185148

>>9183755
usa
>>9183734

>> No.9185151

>>9184712
There are no races, there are only clines

>> No.9185205

>>9175738
it's p useful and probabilities are interesting in their own right
it's actually incredible the amount of things we know about what essentially amounts to "random shit happening" and the level of precision with which we can talk about things like stochastic processes

t. financial math phd student

>> No.9185207

>>9175949
like the other anon said, the very nature of statistics makes it difficult to grasp intuitively
the formal rigouer is a nice way of diffusing that since you can always fall back on "by definition" or "by theorem X," which might be a good pedagogical tool for undergrads

>> No.9185212

>>9175963
this
>>9175995

>> No.9185217

>>9177440
autism alert wee woo wee woo

>> No.9185221

>>9177557
ye

>> No.9185223

>>9179642
in a random sample of pills conducted by the pharmaceutical industry, more pills polled were red than not
it's a very celebrated result in statistics you should be familiar with it

>> No.9185226

>>9177657
please explain the difference
i'm interested but i dont understand

>> No.9185240

>>9179647
congrats
double up in math and remember to take advantage of seattle and do an internship

>> No.9185245

>>9181363
>doing machine learning without understanding stats
hello brainlet

>> No.9185246

>>9184688
>falling for the machine learning meme
>settling for 80% accuracy because your shitty ad-hoc model is to stupid too learn better
uh? lmoa @ u

>> No.9185301

How is Bertsekas "introduction to probability"

>> No.9185319

>>9185207
My disappointment is not because undergrads try and struggle to understand some of the fundamental concepts, it's that they don't try at all. To truly understand statistics you have to work with some math, but these days lower undergrad courses are a requirement for many majors. A good number of students are non-STEM focused and they would rather not work with the basic of math concepts whatsoever, but still have the gall to try for a stats minor.

I think the best way for an undergrad to gain a real understanding of the subject is to take the intermediate undergrad course where they teach change of variables, moment generating functions, and basic convergence in probability and weak convergence, and other fundamental stuff. Most of the aforementioned people never do.