[stevenbedrick]

Yes yes yes, please. I live in Oregon, and this time of year[1] pretty much every day is "chance of showers"- but some days that means "off and on, all day long" and other days that means "one or two good downpours, and decent the rest of the day". Trying to guess which it will be at 8:00 AM is a real pain in the neck. I've actually seriously considered trying to train a classifier to make "ride/don't ride" predictions based on the morning hourly forecast.

The problem is that I know just enough about both weather and machine learning to know that such a simplistic approach probably wouldn't work very well. And, of course, the relative costs of false positives and false negatives are very different (false negative: I miss a good riding day; false positive: I get soaked and have to deal with slippery pavement).