Patrick Andrews (a data analyst and blogger) wrote an interesting post the other day about using Facebook data to predict the actions of your girlfriend.

The concept is simple and obvious to an IT engineer/data analyst, but to a regular Joe you might need a little explanation. It’s a well-known fact that history repeats itself. Things in our world happen in patterns, and while some of these patterns are too elaborate for us to understand, some of them can be nailed down with an impressive degree of accuracy.

By analyzing the data from the past, it’s possible to predict the future. This is one of the fundamental principles for any business decision. Whether you’re inventorying and ordering stock or trading stocks on wall-street the first step is to understand the historical trends with a very important rule of thumb:

If a specific set of circumstances yields a specific result, it’s reasonable to assume re-creating those circumstances will yield a similar result.

Andrews, sensing the doom of his own relationship, took to analyzing Facebook data to try and confirm his foreboding. What he learned is that it is absolutely possible to use historic Facebook data to predict changes in your relationship, most notably if you’re about to be dumped.

Women who are “plugged-in” to Facebook all use it similarly. Facebook is a superficial front, it’s used to showcase the qualities and traits which they like, and they omit any and all they do not. People put all of the best moments they have on Facebook and omit the worst (unless they’re looking for sympathy).

By analyzing the number of posts/pics/tags/statuses and messages to or about you in relation to her total posts, you can get a fairly good idea of just how well liked you are by that girl.

If you can graph the number of interactions against the average relationship interactions graph, you will be able to see trends in your relationship. You’ll have a massive spike during the honeymoon stage which will steadily decrease until it levels off around “happily dating” land.

If your interactions line begins to dip after a long plateau, you can be almost certain you’re headed for a crash.

While Andrews does list some average stats:

  • Interactions in the first year: 13.5
  • Change from year 1 to year 2 (in relationships resulting in marriage): 58%
  • Interactions second year: 22.9
  • Change from year two to three (in relationships resulting in marriage): -24%
  • Interactions year three: 17.5

I wish he would have gone further. I would like to see this developed into an app, along with the ability to scale by percentage of change to avoid skewing the data because you’re with a rampant poster. The actual number of posts is irrelevant; it’s the change in interactions which can be used as a sign.

You can find the original article here: Predicting your ex through data analytics and Facebook.