Mining the Social Web

Geographically Clustering Your Network

With the know-how to access extended LinkedIn profile information, and a working knowledge of common clustering algorithms, all that’s left is to introduce a nice visualization that puts it all together. The next section applies k-means to the problem of clustering your professional contacts and plots them out in Google Earth. The section after it introduces an alternative visualization called a Dorling Cartogram , which is essentially a geographically clustered bubble chart that lets you easily visualize how many of your contacts live in each state. Ironically, it doesn’t explicitly use a clustering algorithm like k-means at all, but it still produces intuitive results that are mapped out in 2D space, and it conveys a semblance of geographic clustering and frequency information.

Mapping Your Professional Network with Google Earth

An interesting exercise in seeing k-means in action is to use it to visualize and cluster your professional LinkedIn network by putting it on a map—or the globe, if you’re a fan of Google Earth. In addition to the insight gained by visualizing how your contacts are spread out, you can analyze clusters by using your contacts, the distinct employers of your contacts, or the distinct metro areas in which your contacts reside as a basis. All three approaches might yield results that are useful for different purposes. Through the LinkedIn API, you can fetch location information that describes the major metropolitan area, ...