Linear Algebra for Machine LearningWatch
I'm in year 12 and I'm self studying linear algebra to use for machine learning projects. I'm learning this through Gilbert Strang's MIT lectures. I was wondering how much of linear algebra I need for machine learning, as the full course seems quite long.
I'm trying to implement linear and logistic regression algorithms.
Its good to understand how to implement stuff like that yourself though ... but it will take longer.
Linear regression is obviously a core part of linear algebra. If you do it efficiently/robustly, you might be implementing some form of SVD (singular valued decomposition). But few people would write it (well) from scratch. You can obviously do some from of simpler (algorithmically) matrix inversion routines which would require significantly less linear algebra.
So its not a simple question, you can get away with some basic matrix vector knowledge, but have to accept the matrix inversion stuff will be "basic", but that may be good enough.
How much data (variables, exemplars) are you expecting to deal with?
I'm expecting I'll use official government data for my first project, which is to do with unemplyment rates. I'm trying to see if this has any connection with stock prices using linear regression algorithm.