Tuesday, November 6, 2012

Election Decision Tree

We just went through a section on decision trees in my Machine Learning course, so when I came across this  really cool interactive election decision tree I was really excited.

http://www.nytimes.com/interactive/2012/11/02/us/politics/paths-to-the-white-house.html

The article doesn't call this a decision tree, but that's exactly what this is.

A decision tree splits all of the possible outcomes based on the data that makes the largest difference in the predicted outcome.  In this case, the outcome of Florida makes the biggest difference.  If Obama wins Florida, Romney must win every other battleground state to win.  Next in importance is Ohio, then North Carolina, and so on.  Each contributing successively less information to our prediction.

Isn't robotics amazing?!  It can even be applied to politics.

No comments:

Post a Comment