Google Correlation: What Does This New Tool Do?

The minds at Google started wondering about the flu.  Does flu activity correlate to actual search activity for the flu and other related search terms?  It does, and Google created Flu Trends to give a prediction of flu activity.  According to Google, they have been complimented by public health officials for providing such helpful, accurate information.  So what does this mean for you?  It means that you can use the Correlate tool to start with a trend and then correlate that data to search terms.

How does it work?  Google used the flu as an example: they compared time series for millions of search terms with actual flu data to get a list of the queries that best correlated with the flu trends.  To use it for your business, whatever type of data you process or have a need to know about, you would enter in the time series (which is just a sequence of data points) you wanted to compare, and Google would create a list of correlated queries.

If you don’t have a time series, or have no idea what a time series is (for those of us who have not majored in statistics), you can simply input the search term, like “flu,” “liposuction,” “car sales,” or whatever.  Google creates the time series, and then goes from there on its own to get your correlated queries.

With Correlate, you can figure out correlations over time, such as Google did with the influenza example, and by state, which is of importance to regional manufacturers and others.  Still in its experimental stage, Google makes a point to remind everyone that correlation is not causation!

To prove this point, the Wall Street Journal used Correlate on several queries.  They found a correlation between the growing federal deficit in the United States and with searches for “nausea remedies.”  The correlation was very high at 0.9605, but even higher for “how to get over a guy.”  Equipment manufacturer Caterpillar was shown to have a high correlation with “video estupidos.”

Correlate doesn’t tell you what causes a trend or why searches for the US deficit would be heavily correlated with nausea (or does it?).  It simply tells you that they are correlated.  What you do with that information is up to you.