Thursday, October 27, 2011

Facebook regression

It turns out that using an exponential function to model the facebook population wasn't the best choice.  Check out the results!

(Can you tell I had fun making this? I love the results of this question!)


  1. It would be interesting to include the carrying capacity of Facebook that is predicted by the logistic regression. It also would be interesting to use the quadratic regression to predict when Facebook will have 0 users.

    I use Facebook to model exponential functions with my high school class (this is the dual credit class at my community college.) I have them count their friends on Facebook (P). Then I have them count the number of friends of friends that are not mutual friends (a). They use the exponential function F(t) = Pa^t to predict the number of people who are t degrees of separation from them. They then set this equal to the population of the Earth to predict the number of degrees of separation between two people. It is always less than six degrees.

    I'm pretty sure that the model is not correct, but it does get the students involved and making interesting predictions. I have them critique the methodology as well.

  2. Thanks for the comment! I actually had someone bring up the "six degrees of separation" idea in my first class, so I may throw this idea at them. What great timing! :)