Wednesday, May 25, 2011

Correlation is not Space Exploration (or is it?)

I was intrigued by Justin Wolfers' post about the new data mining tool, Google Correlate. Apparently, it works well, so I had to give it a try. My data? Weekly advertising revenue from my YouTube channel.

Needless to say, I was confused when I first read the results.

What does the cost of space exploration have to do with my economics YouTube channel? I have a video on cost minimization, but that's not quite space exploration. Of the top 20 correlated series (all with a correlation exceeding 0.826), three seemed tangentially related to my channel's content: fungible goods, assumptions of anova, and critical region statistics.

But, there was something oddly specific, obscure and intense about these keywords. Someone who searches these keywords is looking for knowledge and understanding -- perhaps for a report or an exam.

As an alternative, it is fun to see the flip side. The search terms have an entirely different flavor when we look at what is most strongly negatively correlated with my YouTube revenue (just by multiplying the column of data by -1).

Funtastic. I will have to play more with correlations like these. Expect more in the future.

No comments:

Post a Comment

Please feel free to share your ideas about this post in the open forum. Be mindful that comments in this blog are moderated. Please keep your comments respectful and on point.