Tuesday, March 9, 2010

Day 3 update: the magical land of spring-break campus

Ok, so today went a little differently than expected...

I read in the morning as usual for this week, but did so at a coffee shop in near campus instead of at home, to mix things up and to get some caffeine into the system. Spring break is a marvelous thing - I didn't have to wait in line at the shop, I didn't have to fight for a table, and I didn't have to wear headphones to drown out the usual crowd of students and others. I got a seat near the fire place AND the window. Magical. Walking around campus is a lot more fun too - fewer cars, fewer maniacal bikers trying to run me over, and fewer slow walkers to dodge (yeah, I'm one of those fast-walking types). The only downside is that the library closes early. Ooops. Guess I'll have to wait another day before delving into the catacombs of hierarchy theory.

Mid-day I showed up at one of my reading groups (having not been prepared enough last week to imply upcoming absence). But it was cool - one of the better little interactions I've had with such groups - everyone was tasked with bringing a single slide containing graphs/analysis of data they've been working on. I think I was the only one that rigidly followed the directions, managing to get 5 graphs intelligibly on a powerpoint slide. It was cool though to hear really succinct tidbits about a real diversity of different projects in progress - I feel like I learned something interesting from each one, which is more than I usually expect.

However, the downside to all of this is that, in addition to taking me away from my reclusive reading for a while, it also got me thinking about my current data analysis project again... and I had new ideas that I couldn't wait to try out. It's possible that I spent pretty much all afternoon tinkering with R code. But my new stuff works decently well (still a bit kloodgy) and I think it will save me a lot of time in the coming weeks, as I've been able to generalize a function so that I can give it in short notation the statistical model that I want to fit, and it will go off and do it for me, without my having to write out explicitly what I want each time with minor variations and new names. Pretty exciting. Now if only I could do it more elegantly.... hmmmm

BEWARE: Caution my dear readers; for those of you not already hooked... programming can get really really addictive. It has a way of inspiring the feeling that the breakthrough you've been waiting for is just around the corner, and then holding out success on you for hours or days on end. In this case, I've got a general formula now that lets me combine multiple linear regression and multiple logistic regression on two different distributions combined in a mixture distribution and fit in one fell swoop by maximum likelihood analysis. Fear me.

Oh, and it makes pretty graphs too.

Tomorrow I will strengthen my resolve and return to another day of reading. It's supposed to be rainy too, so I'll be less inclined to frolic around campus pretending that I don't have to share it with anyone. I'm off to delete a few dozen emails before bed.

Theo out.

No comments: