In honor of Valentine's Day, Facebook revealed something very creepy. With the tons of data it has stored from its members, Facebook had a pretty good idea. He decided to use them to predict how long a romantic relationship will last between two ανθρώπων που είναι στο social network.
In a news research conducted by Facebook data scientist Bogdan State, analyzes the relationships of its users Facebook in the US from 2008 to 2011. He found that the most important predictor of whether or not a relationship will last is space where the couple is together.
Ο Bogdan State concluded that “about half of all his relationships Facebook who have survived for more than three months are likely to survive for four years or more ”which seems to make sense. As time goes on, unstable, and casual romances end, leaving a higher percentage in strong and stable relationships.
The study focused exclusively on relationships that lasted at least three months and whose members were at least 23 years old. It also does not include anyone who is referred to as "married". The study did not take into account relationships that end before three months and, of course, those who have not formally declared their relationship to Facebook.
Here is a graph showing the "estimated probability of a relationship surviving a certain duration, for relationships lasting three months or more."
The study also looked at separation combinations, and found that relationships are more likely to end during the summer. It may be the academic calendar year that takes people away from their holiday partner, or it may be the season itself. In fact, during the period of 2011 there was an alert of divorces. " "It could be the recovery of the American economy that happened at the same time," he wonders. State. I doubt.
His tendency Facebook to try to understand and predict human relations begins with its founder. According to 2010's book by David Kirkpatrick "The Facebook Effect" The Mark Zuckerberg, in his spare time, found that “looking at human relationships and ways of communicating he could determine with an accuracy of about 33% when a user was going to have a new relationship.
