Google has been using search data to model flu outbreaks for a number of years. Generally, they look at the number of searches for words related to “flu” (graphs) and look for sharp increases in the number of searches (slopes, derivatives). You can watch a short and elegant video showing the results here. They’ve been tracking this data against the CDC data (graph comparison) and although the CDC data lags behind one to two weeks, it looks like the Google data can accurately predict when a flu outbreak starts (only, in real time). The data follows a nice periodic graph which you should be able to model with a Fourier Series.

Want the data? Google will give you the data (go here).

Perhaps, intrigued by this, you’d like to see if search trends make any other predictions – perhaps you could’ve predicted the recession in January 2008 (instead of 12 months later) by looking at searches for bankruptcy, unemployment, jobs, etc. You could assign a project to your students and have them use Google Trends.

Unfortunately, I’m about to take a sabbatical, so I can’t do this with a class next semester, but I’m thinking that it would be a good project for Honors Calculus in the fall! I think I will have each student come up with a set of terms to track that are related to a topic they are interested in, and then track the data a few times a week every week. When they notice sharp differences in the curve (looking at the slopes), we could look for indications that changes in the world are causing changes in the their topic. Perhaps we will see that they, too, can predict what is about to happen based on sudden changes in search data. Fun! (I almost wish I wasn’t going on sabbatical … almost…)

Possibly Related Posts: