[Python] How do I distinctly label a sparse cluster from a dense cluster in a dataset?

I am trying to figure out a means to group and visualize the dense vs sparse clusters in 2 feature datasets.

When I attempted K-Means, it cut the data in half perpendicular to the trend and was unable to separate the obvious bottom cluster from the above outliers. I then attempted DBSCAN as my method of attack against clusters of varying density however I ended up with this: DBSCAN Clustering

What is the best way in python to differentiate the dense cluster that follows the trend line and the sparse cluster which contains my outliers, and how can I generalize this for multiple 2 feature datasets.

Continue reading...