Monday, November 12, 2012

Ignore Bayes At Your Peril

...are you listening, Karl Rove?

Thanks to the success of Nate Silver in projecting the U.S. Presidential election results, Bayesian statistics have been getting a lot of internet play of late. Bayes' Theorem "bridges the gap between probability and logic" by looking at events not solely in terms of randomness, but in terms of updated known information that can be factored in (probability becomes "conditional" instead of simply random). The idea is simple though the actual application can get rather technical. Here's a basic introductory post to the subject from Grey Matters blog (which includes several additional links and videos):

Plenty more on Bayes' Theorem from a Google search:

...and on YouTube as well:

...And just today 'Mind Your Decisions' blog posted about a simple medical diagnosis problem, which hints at the potential pitfalls of a purely "frequentist" (non-Bayesian) approach to statistics: 

No comments: