Drawing with Google Correlate

This week, Nick Hatch (twitter) showed me the Search by Drawing feature in Google Correlate. My reaction was “That’s going to consume several hours of my weekend” and sure enough, it already has 🙂

Getting started is easy. Go to Google Correlate (you’ll need to sign in with your Google account) then click on the Search by Drawing link on the left side of the page under the Correlate Labs section. With that, you’ll be presented with a blank chart where you can draw a time series of search activity… just draw a line, click the Correlate! button and see what happens. The tool will display your line with a line of activity for search terms. The most correlated search term is displayed initially, but other search terms are presented as well (ranked by decreasing correlation) and you can click on those terms to display their lines.

You might be wondering how could I spend hours drawing lines. Fair enough.

I started thinking about search terms and imagining what a time series for that term would look like, then drew it, then looked at what my line actually matched. For example, try to think of the number of searches over the last several years for Lady Gaga or Charlie the Unicorn, then draw that time series and see the results.

Some of my more interesting attempts:

  • The line I drew that I thought might look like the time series for MusicMatch had a 0.9756 correlation with AltaVista. This was interesting because Yahoo ended up acquiring both of these products.
  • The line I drew trying to guess search activity for FarmVille correlated most highly with Dropbox. This wasn’t too interesting, but the data had something unexpected. As I looked at the time series of search activity for Dropbox, there wasn’t a surprise in the general trend. Dropbox was founded in 2007, so sure enough in 2007 the time series shot up exponentially. Before that, the search activity was perfectly level, except for little squiggles around 2005. Hmmm…
  • I drew my imagined time series for searches on the mortgage banking industry, and my line’s highest correlation was with a bank that was purchased after the 2008 banking crisis. However, well before the banking crisis, there was a sudden and dramatic peak in search activity for that bank. With a little searching, I read that the bank encountered regulatory activity during the period of that peak. So looking at the data prompted me to do a little digging and learn something.

Anyway, if you’re a data geek and haven’t poked around with Google Correlate yet, then you might want to check it out. Enjoy!

About Noel McKinney

Noel McKinney is the administrator of noelmckinney.com
This entry was posted in Data and tagged . Bookmark the permalink.

One Response to Drawing with Google Correlate

  1. Pingback: Labor for the Data | Noel NOT NULL;

Comments are closed.