As a fun project this summer, my boss gave me some work time to use data mining techniques to predict the outcome of "Dancing with the Stars." (I work at a data mining company.) In fact, the goal was to use only the demographic-type statistics -- no dance scores -- to predict who will make it to the Finals. It was really an interesting project, and I got some results that surprised me. For example,
- The relative fame/notoriety of the contestants prior to being on the show does not affect their outcome. That is, fans of the show don't seem to just vote for the contestants they know, but rather for who dances the best.
- Ethnicity also matters very little. The data shows that contestants who are of an ethnic minority fare as well (in fact, a little better) than white contestants.
- Athletes have been rumored to do better than other celebrities, and our investigation confirms that.

I wrote an article with our full results. If you'd like to read it, it is at http://discoverycorpsinc.com/dancing_with_the_stats_article/" target="_blank">Dancing with the StaTs

• RE: Using statistics to predict the outcome of DWTS,Round Robin, 00:44 AM, 10-01-11
• Great Article!,Flowerpower, 12:17 PM, 10-02-11

Their relative fame or notoriety might not factor in that much, but I'd bet that if there were a statistical way that you could factor in WHAT they were famous or notorious for, it would show that the specific reason they were famous or notorious did in some cases affect their results.

Welcome to the Boards Dantastic! I LOVED your article and found it fascinating. I think you have really covered all of the bases with so many variables. While I don't think that Rob Kardashian will last through to the Final 3, I think this is a fantastic tool to perhaps predict 2 of the 3. I think the Soap Opera guy may get in there as well, and I wouldn't discount Derek and Ricki Lake either.

Fascinating stuff! Thanks for sharing!