A succcessful predictor

fivethirtyeight.com

Polling data

http://www.gallup.com/

Weighting the data

http://www.fivethirtyeight.com/2010/06/pollster-ratings-v40-methodology.html

Key idea

To predict X use data related to X

Key idea

To predict player performance use data about player performance

Key idea

To predict movie preferences use data about movie preferences

Key idea

To predict hospitalizations use data about hospitalizations

Not a hard rule

To predict flu outbreaks use Google searches

http://www.google.org/flutrends/

Looser connection = harder prediction

Data properties matter

Unrelated data is the most common mistake

http://www.nejm.org/doi/full/10.1056/NEJMon1211064