This article by Ed Yong in The Scientist is included because it describes a real-world example of a prediction problem in the health/medical sciences domain - training a classifier to predict risk of autism spectrum disorder (ASD) based on genetic markers - as well as including discussion of potential overfitting of the classifier (by training and testing on the same data) as one possible issue, among other factors, by researchers attempting to replicate the study.

http://www.the-scientist.com/?articles.view/articleNo/38030/title/Genetic-Test-for-Autism-Refuted/