Google Correlate lets you see how your data relates to search queries

A while back, Google showed how Influenza outbreaks correlated to searches for flu-related terms with Google Flu Trends. It helped researchers and policy-makers estimate flu activity much sooner than with previous methods. Google Correlate is the evolution of Flu Trends in that now you can correlate search trends with not just flu cases, but with your own data or other search queries.

The above, which you already know about, matches flu cases with searches for “treatment for flu.” Similarly, the search phrase that correlates highest with “Toyota for sale” is “used Hyundai,” as shown below.

You can also see how your data is related geographically. For example, annual rainfall (left) strongly correlates with searches for “disney vacation package.” Although, it looks like distance is a strong factor in the latter, which should be a reminder that correlation is different from causation. Google is careful to point this out in their FAQ and explanation of the tool.

Nevertheless, it’s fun to poke around and sometimes see the non-sensical correlations. For example, the strongest correlation with “flowingdata” is “how to scan a document,” because the growth rates of both seem similar.

There’s also a search by drawing function. You draw a time series, and Correlate finds terms that best match that trend. In the below chart, I drew a line (blue) that had steady growth, but plateaued towards present day.

What weird correlations can you find?

[Google Correlate]

17 Comments