Dean Abbott wrote a pair of posts on a New York Times article about Target predicting if customers are pregnant.
Target, Pregnancy and Predictive Analytics (part 1)
Target, Pregnancy and Predictive Analytics (part 2)
Read both I truly liked his conclusion that models give us the patterns in data but it is up to us to “recognize” the patterns as significant.
BTW, I do wonder what the different is between the New York Times snooping for secrets to sell newspapers versus Target to sell products? If you know, please give a shout!