I enrolled in a class at Stanford. Stats 315A - "Modern Applied Statistics: Learning", first in a sequence of Data Mining courses I'm planning to take. It's not for any degree program, just for sharpening my professional skills. Knowing the techniques of data mining will be extremely helpful for pulling useful insights out of the Test Pilot data set. It will also be a useful set of skills to have for just about anything I do in the future, I think. For example, if I end up leaving software and working in green technology, as I often daydream about, then data mining techniques will surely be useful in teasing out the important relationships from complex ecological systems or finding the best ways to improve energy efficiency of some machine.
The course is all online - the lectures are videos posted to the web, and I download problem sets and email answers back to the professor. I'm learning to use a programming language / statistical manipulation environment called "R". Don't have much to say about it so far other than that it seems pretty challenging.
Stats 315A was an expensive mistake
When the subject of "the expectation value of the p-dimensional matrix product of multivariate gaussian distributions" was broached, my sparing human intellect instantly assumed the most ingratiating posture of surrender imaginable.
Well, not really. But I did realize that the statistics class I signed up for was beyond me. It made me feel really, really stupid.
I got the first homework assignment and spent about a week banging my head against a brick wall. The textbook was too advanced for me so I spent a lot of time searching the internet for definitions of basic terms. This week overlapped with my Brazil trip, so I regrettably spent my day in Rio de Janeiro indoors trying to do statistics homework, instead of going to the beach like a sensible person.
I could do the coding exercises for the class (and in the process I learned a lot of R coding, which is the one useful thing I got out of this) but I just don't understand the probability math. I need to take an easier class where I can learn the mathematical properties of probability distributions and expectation values and conditional probabilities and learn how to manipulate them.
I dropped the class, but not in time to get my money back. I'm not going to complain about that, because it was totally my own fault. It was an act of pure hubris to assume that I could do a Stanford 300-level statistics course without taking the prerequisites first and I compounded folly upon folly by taking so long to figure that out and missing the refund deadline.
I still want to learn data mining. I just have to learn to crawl before I learn to walk.
Guess who's applying to grad school again
Ever since my falling out with the software industry, I've been talking about maybe going back to grad school to learn some new skills. Like either getting back into the hard sciences (which I regret abandoning) or learning some engineering skills (real engineering, not this slipshod circus we call software "engineering").
Especially if I want to go into green energy tech, since my most recent attempt to break into that field left me feeling that my current skillset is just pigeonholing me in the role of web-monkey.
Stanford and Berkeley both have graduate programs in green energy (i.e. in the weird intersection of policy, engineering, science, and business that it will take to make a dent in humanity's fossil fuel addiction). They are both top-tier schools. Stanford is practically across the street; Berkeley is up north, but not so far that Sushu would have to leave her school. We could find a compromise address that allows us both to commute.
All autumn long I'd been running my mouth off about applying to grad school but not actually doing anything about it, because I procrastinate like a champ, especially when it comes to fractally tedious tasks like setting up applications.
Then Saturday morning Sushu finally got sick of my procrastination and kicked my ass into gear. We looked at application deadlines and discovered that I had less than a week to apply to the Energy Resources Group at Berkeley. There's a Dec. 7 deadline to apply for Fall 2013.
After a mad scramble for transcripts and recommendation letters it looks like I might actually make the deadline.
I have to take the GRE again. Last time I took it was like 1998 and they only keep your scores for five years, so I need to do it over. I've got an appointment for Thursday.
Feels weird to be preparing for a test again after so many years away from academia. I'm not worried, though. In 1998 I was nervous because I thought the GRE would, I dunno, measure my worth as a human being or something. Now I see it as just a bureaucratic obstacle that doesn't really mean anything. Don't sweat it, just get it over with and score whatever I score.
I looked at some sample GRE questions this morning and most of them are insultingly easy. Like, 6th graders should be able to answer most of these math questions, no offense to 6th graders.
Many of the grammar questions, meanwhile, are total bullshit. They're not looking for whether you can communicate clearly in the English language, they're looking for class markers, i.e. "prove you write in the dialect of an educated upper-class white American and not any other English dialect". I could rant about prescriptivist grammar but that would be a whole other blog post.
Even if I get accepted, that doesn't mean for sure I'm going back to school. I'll have to weigh it against whatever other opportunities I have before me in fall of 2013. But applying can't hurt.