For the second year running our Data Mining Section has offered a competition to attract friendly interaction and have some fun “Digging thru data”. Teams solve or “score” the stated predictive modeling problem with data provided in collaboration with Health Care Intelligence . Focusing on health care quality , contestants were given access to Data sets that were composed of sequences of hospital discharge data.
The challenge had two parts.
Task 1: Model the transfer decision (from one hospital to another)
Task 2: Predict mortality.
This year we had 250 teams register for the contest. 26 teams submitted their results in the given 4 week deadline. Certificates were presented to the winners during the Saturday workshop and three of the finalists will share their analysis in Tuesdays Session TD08 in convention center room 24C at 4:30pm. You can read about the problem description and find the WINNERS listed on this website. http://www.informsdmcontest2009.org/
I’d like give my personal call out of “ Thanks” to Nick Street of Iowa State – our first organizer of the data mining contest last year – because I credit him with the recruitment of a brilliant new member to our INFORMS community. Claudia Perlich, of IBM, brings years of active involvement with the ACM society (a multiyear KDD cup winner) – but had never considered joining or attending an INFORMS event — until Nick Street’s contest. Here in San Diego she is an active volunteer, you will find her chairing TD08 , as well as presenting with her team in the reprise of the Edelman finalist session TC26. Do stop by and welcome her!
Do you have friends who are not yet INFORM members – perhaps in other professional societies? Please encourage them to share their analytical contributions – and join INFORMS! It’s the multi disciplinary aspect to our profession that adds value and gives us strategic perspective to addressing critical issues such as National Health Care today.
Copyright © 2009, INFORMS | Institute for Operations Research and the Management Sciences | Feedback.