so this contest is done:
http://www.heritagehealthprize.com/
initially i was excited to hear them offering 3 million dollars to create an algorithm to predict who will be readmitted and how long, which has good implications in healthcare
however, i was discussing with some people about this competition and they told me that the models the winners created were no good. in other words, it can only predict marginally better than taking an average of who was readmitted.
im new to machine learning, what can anyone shed some insight? or discuss? is it true the model the winning team created cannot be used in the real world? what does the "0.46" they got mean in eli5 terms?
[link][comment]