Ask HN: Developer with little mathematical knowledge, how do I fix this?

5 points by bennysomething ↗ HN
I've been a developer for about 15 but I don't have any maths education beyond 16 years old. So no calculas (I'm in the UK). I've always known that this was a limiting factor for me and greatly reduces what areas I can work in. I'm nearly 40 and I want to fix this. Is there a single book or maybe series of books that could work through that get me to at least doing basic calculus? My basic algebra probably needs quite a bit of work first. Maybe there's an online course I could do? Khan academy maybe?

Any thoughts or ideas would be really appreciated.

Edit: I'm also really scared of "getting found out". It greatly adds to my feeling of imposter syndrome.

Also I really want something that I can't just work through, a structured path so I can see my progress and not constantly feel like I should be studying some other area all the time.

10 comments

[ 0.20 ms ] story [ 38.7 ms ] thread
Machine learning is inherently mathematical. Maybe people can recommend machine learning books that teach the calculus and linear algebra needed "along the way". I like the books by Aurelien Geron and Joel Grus but would have to review them to see what mathematical background they assume.
Thanks but I think I need to go even further , to high school maths.
Calculus isn’t used too much in the software biz except for neural networks and such.

Linear algebra is important, as is combinatorics. Look at the required classes for a CS (or math) major, subtract the ones required for a physics major and I’d say that is your curriculum.

Thanks, thing is I don't even feel I'm ready for CS degree level maths yet, I need to go back further.
You may be more competent and motivated than I was. However, I went back to college classes to learn math, supplemented with a healthy dose of Khan Academy.
Do you mean like a sort of night school course? What level did it take you too? Ready for a numerate degree?
I left work and became a full time student. Took through real analysis in the math dept, and ended up getting a masters in statistics. Now work full time as an ML researcher.
That's impressive, how old where you when you became a full time student and what were you working as? (If you don't mind me asking).
I've found it really difficult to learn new topics like this comprehensively, and I'm in a similar age range, so I feel the challenge you describe.

My perhaps old-fashioned view is that taking some kind of course -- online or otherwise -- that involves a prepared curriculum, work assignments, and examinations is probably the way to go.

Piecemeal learning may be effective under some circumstances, but for a large field that involves inter-related concepts, my guess is that it's hard to build a solid understanding without having stepped through a carefully-designed set of lessons that highlight and anticipate the connections in advance.

I'd also expect that finding the time to take that kind of course and commit sufficient energy to it may be a challenge. Could you ask your employer to see whether they'd be supportive of it? (perhaps with career and role development in mind?)

Another potential factor to consider: having a supportive social group during education can be very effective, I think. There's no magic solution for that; it'll involve a bit of outreach and trying to find those people to learn with.

That doesn't provide any neat answers or links for you, I'm sorry about that; but I'm curious to follow this thread and the responses, and to let you know that you're not alone. Good luck!

Thanks, I completely agree with you on the defined course thing. I'd like a defined path, I hate trying to learn about any topic from random webpages with links scattered throughout.

I considered going back and doing the sort of last year maths exam from UK high school, A-level or Scottish higher for this reason. I like the accountability and enforced deadlines. And that the curriculum has been designed properly.

Either that or work through khan academy.