# Wrapping Up

Between the graph modularity chapter and this chapter on outlier detection, you've been exposed to the power of analyzing a dataset by “graphing” your data, that is, assigning distances and edges between your observations.

Although in the clustering chapters, you mined groups of related points for insights, here you mined the data for points outside of communities. You saw the power of something as simple as indegree to demonstrate who's influential and who's isolated.

For more on outlier detection, check out the 2010 survey put together by Kriegel, Kroger, and Zimek at `http://www.siam.org/meetings/sdm10/tutorial3.pdf`

for the 2010 SIAM conference. All the techniques in this chapter show up there along with a number of others.

Note that these techniques don't require any kind of arbitrarily long-running process the way optimization models might. There are a finite number of steps to get LOFs, so this kind of thing can be coded in production...