That's a good question. Feature selection is a large field of research and is a bit too broad for me to summarize in an abbreviated fashion. I would look into "model selection", specifically into scores of models that weigh both complexity (the number of variables) and goodness of fit. A good score to look into first is the Bayesian information criterion (BIC) which is used, for instance, in model selection in neuroscience. http://en.wikipedia.org/wiki/Bayesian_information_criterion
One thing you might want to try is cross-validation (http://en.wikipedia.org/wiki/Cross-validation_%28statistics%...). Cross-validation should help you determine if your model is overfitting, as it will perform significantly better on its training set than on the left out data.
A few years ago some SVM code was contributed to the Mahout project, but as of yet, it still doesn't appear to have a working implementation. It seems one can tweak existing Mahout functions a bit in order to accomplish the same sort of thing, but Mike went ahead and started working on an SVM implementation when he initially discovered it wasn't fully implemented. Given a package like sklearn (part of the Python scipy package), it's not so hard to implement a scheme similar to the one described in the blog once you know what to do.
No, the Times episode was clearly the result of an ill-considered and ill-timed system update that went south. Not that the Times isn't sometimes hacked, but yesterday's episode had a pedestrian explanation.
There are tons of distractions within the car - changing the radio, changing the song on your ipod, looking through your purse, eating some food - texting is only adding to the list. THe problem is there is too much for us to do and until all of those things get taken away accidents will continue to happen.
Obviously texting is extremely bad because of the length it takes to read/write a message - but other than cars and phones having a sync to shut down phones within the car - there really isn't an alternative that will be able to stop this problem.
I applaud you for not drinking around her - but you should have seen how long you could go without a drink. She knows she cannot and ultimately has forced will power behind her decision, but you should have seen if you could have done the same. It's so easy after a stressful day to have 1 beer or to wind down with something stronger - but I do believe it shows a lot of character for someone who's willing to 'suffer' with their significant other to show support.
I couldn't imagine not fitting into a company - it would be miserable past a certain point and when you get to that point everyone knows your work and effort start to diminish and most of the time it's definitely noticeable.
One thing you might want to try is cross-validation (http://en.wikipedia.org/wiki/Cross-validation_%28statistics%...). Cross-validation should help you determine if your model is overfitting, as it will perform significantly better on its training set than on the left out data.