Paul Butler, an intern on Facebook’s data infrastructure engineering team, was interested in visualizing the "locality of friendship". Luckily, he has some great data to work with: Facebook's social network of the friendships between its 500 million members. But visualizing that much data can be a challenge in its own right -- it takes skill to draw meaning from what could easily be an incomprehensible mess of data. After drawing a sample of 10 million friend pairs from the Hive interface to Facebook's Hadoop-based database, Paul set to using R to solve this visualization problem.
As Paul describes in a post on his Facebook page, initial attempts to visualize the data resulted in a "big white blob" roughly resembling the outline of the continents. But when Paul switched from plotting every friend pair to instead plotting every city pair with a great-circle line whose transparency was determined by the number of friend-pairs in those cities, something beautiful emerges: a clear image of the world, with friendship bonds flowing between the continents:
| This post is originally from Revolutions; click here the original post. |
No comments:
Post a Comment