Tfw I've found a machine learning self study guide on google

>not making a working prototype and seek capital investors or a kickstarter
>but I need a degree, I was told I need one
facebook zuckerkike didn't even finish college, nor bill gates nor the guy who made apple.

capital investors only care if your idea can make them money.

>.t neet
Running your own business consumes more time, energy and money than being a wagecuck. You are truly the neetest of neets.

>not making a patreon of your AI bot waifu for r9k

>user finds machine learning guide
>haha wow this is awesome
>realizes there are 2-4 years of math prereqs
>promptly deletes site from favorites

HR stacies will block for from even getting an interview if you dont have a line on your resume stating your degree and university

>the thinks math for 18 year old kids is hard

>another wagecuck

they were all rich

Notch made 2 billions of a java game.

unless you're a genius learning calculus and linear algebra will take a lot of time

> Strawmaning the need for a degree
> Thinking you just need a working prototype
> Thinking kickstarter is still relevant
> Thinking all those people were not rich and resourceful
> the guy who made apple

You clearly know shit about what you're talkin' about.

>literally high school algebra for vectors and matrixes is hard
>first year calculus is hard
this is why /sci/ makes fun of CS.

Notch made two billions of a java game and he was just a fat guy making a game for reddit.

Your cuckery is so far beyond Notch's you are not the right amount of autistic to make a successful game for incels nor chad enough to steal a great idea and market it like Gates, Jobs or Lizardberg.

kys.

>but muh degree says Jow Forums
stop your selective hearing, Jow Forums has always said degrees are only worthwhile for connections and a foot in the door for corporate hr staff that don't know what they're doing

I only need to make a waifu bot for incels and make a patreon.

Let us know when you get a job doing ML

Sure you will

both zuckerberg and gates came from rich families, went to private schools, at least in bills case was very talented, both went to harvard, both dropped out to pursue their business ideas after getting all the contacts they needed at university and were safe in the knowledge that they would have the funds to return to harvard if it didn't work out, they are not good rolemodels

notch was a barely competent 32 year old boomer making flash tier games for pre-smartphone mobiles before he came up with the idea for minecraft, his previous game, wurm online, was an incredibly niche grindfest that ran like ass because again he was barely competent
I don't think university is necessary to succeed in cs but you said it yourself, he was 30 years old and unable to progress beyond java and was making java games for reddit (it was more like twitter/his blog at the time), if this is your aspiration then may god have mercy on your soul

He's in Mensa. My IQ isn't over 132 sorry.

I'll laugh when you try to actually apply things or go for jobs

OP, let's start a company. We'll come up with a prototype in three months. Applied to a high value domain that's struggling with the digital transformation stuff. Execs are in panic, shareholders are concerned. We'll demo something that makes them happy. Sign them up. Then in six months we'll get seed funding, then Series A by end of 2019. Are you game?

So you found a self-study guide, and from that you conclude that it is equivalent to a degree? How did you reach that conclusion? Not saying you're wrong, I just don't understand how you have enough information to draw that conclusion.

Personally I think you can self-study a lot of this stuff, the quality of online courses is very high. But it's hard to be proficient/fluent in math (and probability/stats) without some college courses. This stuff is harder to self study because it's difficult and takes motivation, which college courses provide, and also it's hard to self-correct your mistakes.

Degrees (but really certs) is what gets you hired.

>self study
>reading a guide to do so
just do it bro

OP, do you need help with this project? I have some skills with game dev.

This is true I think at big companies with HR monkeys that check resume vs list with checkboxes etc, but you can do pretty much anything on connections/experience.

What? Not really.

it's not that anyone in the tech industry really respects a CS degree and expects any recent 4-year grad to know anything about programming

It's more like a filter we have to disgruntingly put on to weed people out, even though 98% of people who apply with a bachelors are still retarded

I don't know why westerners think that probabilities, calculus or matrixes is difficult. In post-soviet countries this shit is taught at school to the level that is taught in Western universities.

t. Graduate of post-soviet country school studying in Western University STEM

ok but you did learn it in school, you didn't self-study right?

>doesn't realize he needs a graduate degree and 2 years experience to get a machine learning related job
LOL

Attached: unemployment.png (500x384, 311K)

Nope, didn't self-study. My question is why people in Europe/America study such basic math in secondary-high schools

It was different decades ago (speaking for Western Europe). Then we became more diverse and inclusive. Think it started to go down in the late 1980s. Probably 10-15 years earlier in the US from what I read.

CVs without a university education automatically get thrown in the bin

t. manager at a big MSP in Australia

But that's nothing to do with what I wrote in , my point was math and stats are hard to self-study. And I maintain they are hard topics, even if you learn them in high school.

quora.com/Is-Russian-mathematical-education-among-the-best-in-the-world has a nice answer to why (ex)-soviet nations have better math education.

I work in the field and you won't get anywhere near employed doing legitimate machine learning without an MS in math/stats/cs at a minimum.
You may get a job as a Hadoop monkey though, if you're lucky.

I work for a national ISP / MSP. I do "Business development", which is mostly building SAAS platforms.

If CS students learned something less stupid than Java, I'd be more interested in hiring them. Java isn't difficult enough to separate incompetents from useful developers the same way C, Pascal, or any Lisp dialects do.

Would you hire me with Haskell? Elixir as well, if needed.

So if I'm a math brainlet data science isn't for me then

no it's not

>You may get a job as a Hadoop monkey though, if you're lucky.
What about an Ms in Econ?

I am almost done, I could add another year and get stats since their is so much overlap.

Not if you want to do it properly. But you could apply as data plumber. Literally say this. They'll understand and love you for it.

>MSP
Yeah but at companies that require actual technical skill nobody cares as long as you're good.

Strongly recommend the stats. I went into ML after a PhD in econ, but my PhD was mostly stats anyway.

Sounds like a well paid but soul sucking job

How much overlap is there with regression techniques? (besides logistic)

Should I actually spend time truely ingraining study design and everything else related to causality? I do data analysis right now, plan to try to jump to data science after my MS. But I have yet to see anyone ask me if I can consider my results causal or if my standard errors are robust.

Since you're not doing proper data science anyway, you could be close to those who do. And they'll gladly offload this unwanted work (90% of data science is plumbing) to you. And you'd learn certainly something. Perhaps find an alternative route to do data science without going the formal way.

>How much overlap is there with regression techniques? (besides logistic)
Very little. There is nothing in ML like instrumental variables, 2sls etc. There are things like hierarchical logistic regression but not close enough that it's useful.

>Should I actually spend time truely ingraining study design and everything else related to causality?
No. The general concepts of causality are good to have at the back of your mind, it gives you a slight edge over pure stats/ML people. But the specific techniques and theory related to causality are not useful in ML or data science.
>I do data analysis right now, plan to try to jump to data science after my MS. But I have yet to see anyone ask me if I can consider my results causal or if my standard errors are robust.
Yup, these things don't typically come up on ML. E.g. if I train an ML model, I never compute standard errors on the coefficients/weights because I don't care about their actual values.

I just think this is bad advice, you are basically telling him to do something he knows he is bad at, on the basis that you claim other people doing the job are bad at it. Even if someone bad at math can get by in data science, why do this when most other tech jobs don't need nearly as much math.

>What about an Ms in Econ?
Honestly, probably not, unless it's some specialist role doing ML applied to econ or finance or something.

The thing is that the job market is a little overrated. A vast majority of the jobs out there (like literally ~90%) that are marketed as "data scientists" are actually for Hadoop monkeys, data plumbing, etc. which is basically a glorified DB admin. The people doing cool shit, like making prototypes, testing models, etc. are generally PhD level and usually their job is over early in the project.

That's at least been my experience in the field.

Data plumbing isn't much about math/stat. Many data are unstructured, have issues (it's more about having common sense to check their validity) etc.
And with learning by doing he could gain insights into DS aspects and apply to a domain he's familiar with, grow into the whole thing. It's still not DS in the general sense that he could work on other domains then, but it would be perhaps something he likes and can do. For a narrow domain and toolchain you don't need to have the whole knowledge. I learned that when I started to do modeling and simulation. All the great knowledge is wonderful, but once you're stuck with a niche, it's very narrow and always the same. And perhaps he enjoys data plumbing. Like others prefer debugging over programming.

Hell, I dont care what I am actually doing, as long as I can stop reducing all my results into excel.

fair enough

i don't know what all this crap about it being impossible to self-study math is about.

i took calculus in college and literally my entire knowledge of the subject came from reading a textbook i'd bought at a thrift store (the one they demanded we buy was completely useless) and doing every single problem from every single chapter.

as far as i can tell most people's experience of learning math is about the same -- a lot of them don't even attend lectures -- you literally just read the textbook and do the problems until you can consistently produce correct answers.

the reason nobody self-studies calculus isn't because you need to be a genius, it's because it is almost completely useless and not necessary outside a very narrow range of professional pursuits

t. have been employed as a mechanical design engineer and a software developer without using calculus once for anything ever

data science and machine learning literally run on calculus.

but not run-of-the-mill calculus with integrals and whatnot
it's some janky-ass algorithm-calculus

Hurr what's a gradient

Attached: 700.jpg (200x313, 10K)