NOAA’s National Hurricane Center had a huge win last week when it correctly predicted Hurricane Sandy almost 1 week in advance. Sandy was not an easy storm to predict but the use of higher resolution models (NOAA moved from 9 km to 3 km resolution in June), improved extrapolation between real-world data points, and improved modeling allowed this impressive result. In addition, last year for the first time, the National Weather Service developed the capability to input real time measurements from its Hurricane Hunter airplanes into its hurricane simulations.[i]
The second big data win was Barack Obama’s data driven presidential campaign. As disclosed by Time Magazine[ii] the day after the election, Jim Messina created an analytics department 5x as large as the 2008 operation and used techniques similar to that used by sophisticated retail chains. Through their data collection and analysis they were able to predict who in their database to hit up for donations and just how to do it. And they were able to simulate the election results on a state-by-state basis and try out various strategies to bring out the voters. “We ran the election 66,000 times every night,” said a senior official, according to the article.
Just like Billy Bean[iii] moved the world of baseball into the world of data analytics, so too will Obama’s campaign this year change the way future election campaigns are run. If they are able to accurately predict how we’re going to vote, can we please skip all the robo calls and advertising next time around and go right to the voting?
Director of IBM’s Analytics Solution Center
[iii] Read the book Moneyball by Michael Lewis
Poll of the Week
Could your inbox use a little more awesome?
Sign up to get a daily dose of awesome gov-focused resources, trainings, blogs and articles to help you do you job better.