Black_Swan:
blackomen:
Black_Swan:
blackomen:
Not too hard, plenty of ppl with liberal arts backgrounds are doing ML.
Take a look at course.fast.ai and see if you still think it’s difficult. It’s still not straightforward and easy but anyone who can tackle the CFA should have the intelligence, diligence and even math knowledge to understand this stuff.
So my view on that is that those skills will quickly devalue and become commodities. Everyone I talk to, including professors at programs all say the same thing. There’s a flood of people rushing into this, saying data science with limited real math background that are basically running third party ML and AI programs and using a little python and R. Which is good, but it’s a low threshold and supply is on its way to a parabolic ramp up. So most people have said a more math centric core curriculum will have more durable value.
Good point, but do you really want to be in this space or are you making excuses? While learning ML still isn’t a cakewalk, I’d say it’s comparable in difficulty to the CFA rather than, say, rocket science or a Math PhD. Basically doable by most people of average to slightly above average intelligence with some hard work.
How do you get that I’m making excuses from this? I built out said math skills and am pursuing a masters in the more quantitative side of the field. My point was that the soft skill faux data science people that don’t have a solid mathematical background are in for a rude awakening. I’m walking around at the university and there are posters for data science, every school in the area has added data science masters and PhD’s in the last two years (basically comp sci pretending to be math majors), almost every engineering and math student I talk to is tacking on the equivalent of a data science minor and even the professors in the math department are telling me there’s going to be major oversupply of “math lite” practitioners that know some python and R and are calling it data science. I’ve had similar conversations with people in the field and recruiters saying they’re having to be more stringent about the math backgrounds since it’s become a catch all buzzword. Knowing ML or python is going to be the equivalent of excel proficiency in this field in 10 years.
My central point is I’m picking up the skills, but defensively rather than under an illusion of long term value boost and that the core math competency is not something to skip out on.
Seriously, who’s gonna have time to master all of this AND the CFA on top of it all? Think about it. Even if you speculate that everyone needs to know the latest and greatest in so many subjects under the sun to succeed in the future, please do a reality check: how many people will actually be able to master it all in a short time? But the silver lining is if you can, then you will still stand out from the competition since I doubt many people will be of the caliber required to master ML on top of the CFA.
rawraw
April 6, 2019, 2:39pm
#22
What BS is doing is way more valuable than CFA IMO. BS how much do you pay for the courses? I started the Coursera linear algebra course with the leading researcher but had to stop because of moving and switching jobs. Hoping to get back into it and now I live near a lot of colleges, so considering doing it your route
Yeah, basically if you have projects and a portfolio you can point to coupled with CFA or experience similar to it then you’ll be in a good place.
Sales skills are better compensated, though.
blackomen:
Black_Swan:
blackomen:
Black_Swan:
blackomen:
Not too hard, plenty of ppl with liberal arts backgrounds are doing ML.
Take a look at course.fast.ai and see if you still think it’s difficult. It’s still not straightforward and easy but anyone who can tackle the CFA should have the intelligence, diligence and even math knowledge to understand this stuff.
So my view on that is that those skills will quickly devalue and become commodities. Everyone I talk to, including professors at programs all say the same thing. There’s a flood of people rushing into this, saying data science with limited real math background that are basically running third party ML and AI programs and using a little python and R. Which is good, but it’s a low threshold and supply is on its way to a parabolic ramp up. So most people have said a more math centric core curriculum will have more durable value.
Good point, but do you really want to be in this space or are you making excuses? While learning ML still isn’t a cakewalk, I’d say it’s comparable in difficulty to the CFA rather than, say, rocket science or a Math PhD. Basically doable by most people of average to slightly above average intelligence with some hard work.
How do you get that I’m making excuses from this? I built out said math skills and am pursuing a masters in the more quantitative side of the field. My point was that the soft skill faux data science people that don’t have a solid mathematical background are in for a rude awakening. I’m walking around at the university and there are posters for data science, every school in the area has added data science masters and PhD’s in the last two years (basically comp sci pretending to be math majors), almost every engineering and math student I talk to is tacking on the equivalent of a data science minor and even the professors in the math department are telling me there’s going to be major oversupply of “math lite” practitioners that know some python and R and are calling it data science. I’ve had similar conversations with people in the field and recruiters saying they’re having to be more stringent about the math backgrounds since it’s become a catch all buzzword. Knowing ML or python is going to be the equivalent of excel proficiency in this field in 10 years.
My central point is I’m picking up the skills, but defensively rather than under an illusion of long term value boost and that the core math competency is not something to skip out on.
Seriously, who’s gonna have time to master all of this AND the CFA on top of it all? Think about it. Even if you speculate that everyone needs to know the latest and greatest in so many subjects under the sun to succeed in the future, please do a reality check: how many people will actually be able to master it all in a short time? But the silver lining is if you can, then you will still stand out from the competition since I doubt many people will be of the caliber required to master ML on top of the CFA.
Dude, CFA is not that big of a deal. It’s three tests over a year and a half period. The fact that you’re acting like it’s a crowning achievement says a lot about the reality of how easy it is to pick up ML. Besides, if you bothered to read, you’d see that most new undergrads and MBA’s are picking up this skill in their coursework, it’s becoming standard.
rawraw:
What BS is doing is way more valuable than CFA IMO. BS how much do you pay for the courses? I started the Coursera linear algebra course with the leading researcher but had to stop because of moving and switching jobs. Hoping to get back into it and now I live near a lot of colleges, so considering doing it your route
So I think current rate is around $4-5k per course, but my company covers everything.
Gilbert Strang is the man for linear algebra.
PHd or bust. Well, maybe solid MS can also cut it, otherwise you will just be using some one elses libraries. Never got into ML, but all math courses i took only allowed me to understand regression and basic risk models like arch/garch. Prob linear algebra 1/2, ode, prob with calculus should be pre-req to masters in data science.
Ome thing is clear, most people who just learn puthon dont know how to do software engineering. I guess it is good for me
Black_Swan:
blackomen:
Black_Swan:
blackomen:
Not too hard, plenty of ppl with liberal arts backgrounds are doing ML.
Take a look at course.fast.ai and see if you still think it’s difficult. It’s still not straightforward and easy but anyone who can tackle the CFA should have the intelligence, diligence and even math knowledge to understand this stuff.
So my view on that is that those skills will quickly devalue and become commodities. Everyone I talk to, including professors at programs all say the same thing. There’s a flood of people rushing into this, saying data science with limited real math background that are basically running third party ML and AI programs and using a little python and R. Which is good, but it’s a low threshold and supply is on its way to a parabolic ramp up. So most people have said a more math centric core curriculum will have more durable value.
Good point, but do you really want to be in this space or are you making excuses? While learning ML still isn’t a cakewalk, I’d say it’s comparable in difficulty to the CFA rather than, say, rocket science or a Math PhD. Basically doable by most people of average to slightly above average intelligence with some hard work.
How do you get that I’m making excuses from this? I built out said math skills and am pursuing a masters in the more quantitative side of the field. My point was that the soft skill faux data science people that don’t have a solid mathematical background are in for a rude awakening. I’m walking around at the university and there are posters for data science, every school in the area has added data science masters and PhD’s in the last two years (basically comp sci pretending to be math majors), almost every engineering and math student I talk to is tacking on the equivalent of a data science minor and even the professors in the math department are telling me there’s going to be major oversupply of “math lite” practitioners that know some python and R and are calling it data science. I’ve had similar conversations with people in the field and recruiters saying they’re having to be more stringent about the math backgrounds since it’s become a catch all buzzword. Knowing ML or python is going to be the equivalent of excel proficiency in this field in 10 years.
My central point is I’m picking up the skills, but defensively rather than under an illusion of long term value boost and that the core math competency is not something to skip out on.
lets just clearly differentiate between mechanical mat knowledge and actual ability to do analysis.
Yeah, that’s usually picked up in numerical analysis coursework with a grad degree.
Damn I’m pretty advanced. I took algebra in 7th grade! Was in the gate program! It stands for gifted and talented! Ya feel me
PHd or bust. Well, maybe solid MS can also cut it, otherwise you will just be using some one elses libraries. Never got into ML, but all math courses i took only allowed me to understand regression and basic risk models like arch/garch. Prob linear algebra 1/2, ode, prob with calculus should be pre-req to masters in data science.
Ome thing is clear, most people who just learn puthon dont know how to do software engineering. I guess it is good for me
Agreed. There are plenty of useful ML implementations on Github but most of them are sloppily written, even if the code runs well. Makes it very difficult to maintain without some serious refactoring.
https://www.youtube.com/watch?v=IhsXrjdiMaA
you cut me deep, bitch,
cut me like surgery!
hei.so
April 9, 2019, 1:36pm
#35
blackomen:
Good point, but do you really want to be in this space or are you making excuses? While learning ML still isn’t a cakewalk, I’d say it’s comparable in difficulty to the CFA rather than, say, rocket science or a Math PhD. Basically doable by most people of average to slightly above average intelligence with some hard work.
I couldn’t disagree more. The CFA exams simply don’t have complicated concepts or it doesn’t go into depth into those said concepts. High school calculus is more difficult than the CFA. Unless they change the curriculum to include something like deriving the Black-Scholes formula…
I would say having a strong math background is essential to doing well in ML: linear algebra, probability, calculus, algorithms, etc. I’m learning python right now for some simple data analysis but I don’t think I have the ability to learn about multivariate calculus or combinatorial optimization. I know programming won’t help unless I can apply it. Because programming isn’t about programming, it’s about solving problems.
For those of you who are interested, MIT offers all of their courses online for free.
https://ocw.mit.edu/index.htm
If you think your math background is too weak, I urge you to try http://course.fast.ai instead
hei.so:
blackomen:
Good point, but do you really want to be in this space or are you making excuses? While learning ML still isn’t a cakewalk, I’d say it’s comparable in difficulty to the CFA rather than, say, rocket science or a Math PhD. Basically doable by most people of average to slightly above average intelligence with some hard work.
I couldn’t disagree more. The CFA exams simply don’t have complicated concepts or it doesn’t go into depth into those said concepts. High school calculus is more difficult than the CFA. Unless they change the curriculum to include something like deriving the Black-Scholes formula…
I would say having a strong math background is essential to doing well in ML: linear algebra, probability, calculus, algorithms, etc. I’m learning python right now for some simple data analysis but I don’t think I have the ability to learn about multivariate calculus or combinatorial optimization. I know programming won’t help unless I can apply it. Because programming isn’t about programming, it’s about solving problems.
For those of you who are interested, MIT offers all of their courses online for free.
https://ocw.mit.edu/index.htm
Think simply learning the available ml libraries has a much better roi. I mean 300hours on tensorflow and you are a magician on it VS semesters on math with no direct practical use, not a difficult choice
Wait how do you get to no direct practical use? The only people saying that are people that don’t seem to understand the applications, it shows up in simple programing and analysis all the time. Anywhoo, owing to its simplicity, the ROI is great on learning ML for a year or two, but I see little lasting moat. To basic to pick up, like saying learn excel in 1999 to differentiate yourself over a career. Also, definitely not an either or thing.
Lol most bsds don’t even know how to use Microsoft office.
you will be automated away