Implement machine learning data clustering

[mdibello]

Implement a clustering algorithm (either k-means or x-means) to group similar data and display the results in a 2-D space. For the use-case, implement this machine learning functionality for the pop culture project.

Create some kind of interface where you specify the independent/predicting variables vs. the dependent/measured ones, and maybe the number of clusters (if k-means; x-means tries to figure this automatically). Then it runs the clustering algorithm.

The clusters are in a high dimensional space, but maybe somehow they can be projected into a 2-D view so you can visualize them (see how many points are in each; how dispersed they are). Then you can mouse over them and see the aggregate characteristics of the individuals who make up the clusters.

(This feature has been submitted on behalf of @fgmart)