Thursday, June 16, 2011

Some Statistical Modeling Inspiration

The other day, I was checking weather.com when I noticed that "Chance of Rain" was a hyperlink. See:

I was curious, so I clicked on it and read their explanation:

The 'Chance of Precip' describes the likelihood of precipitation (maybe not even measurable) in your forecast area. The chance is based upon a series of 100 instances of identical weather conditions.

For example, a 40% Chance of Precip means that precipitation occurred 40 out of 100 days with the same type of conditions expected in your area.

In a sense, to predict whether or not it's going to rain you don't really need to know anything about meteorology. You just need enough past data. And bam, you've got your answer.

I've learned this lesson before, but every time I see statistics used to solve a problem that a brute force solution would fail on, I'm in awe.

A promise to myself: keep my eyes open for how this same pattern can be applied to future projects and challenges.

No comments:

Post a Comment