It all started with the flu. In 2008, we found that the activity of certain search terms are good indicators of actual flu activity. Based on this finding, we launched Google Flu Trends to provide timely estimates of flu activity in 28 countries. Since then, we’ve seen a number of other researchers—including our very own—use search activity data to estimate other real world activities.
However, tools that provide access to search data, such as Google Trends or Google Insights for Search, weren’t designed with this type of research in mind. Those systems allow you to enter a search term and see the trend; but researchers told us they want to enter the trend of some real world activity and see which search terms best match that trend. In other words, they wanted a system that was like Google Trends but in reverse.
This is now possible with Google Correlate, which we’re launching today on Google Labs. Using Correlate, you can upload your own data series and see a list of search terms whose popularity best corresponds with that real world trend. In the example below, we uploaded official flu activity data from the U.S. CDC over the last several years and found that people search for terms like [cold or flu] in a similar pattern to actual flu rates.
Google has also published a white paper (PDF) about the methodology Google Correlate uses.
Users can also search and access correlations using Google U.S data and time series data.
Very Cool! Search By Drawing
“Draw an interesting curve, then click ‘Correlate!’ to find query terms whose popularity over time matches the shape you drew.”