Of Interest

About Blog info

Posts Humor, longform, thoughts

Projects Cool stuff

Contact Me

Twitter I only retweet

Email Prefer job offers

Social Network Graph

April 13, 2017

The visualization probably doesn’t look very good on a mobile device. Here it is by itself. Drag and hover to interact with the visualization.

Yielded Students: Preadmit Social Network

I recently finished the D3 social network graph that I started a couple months ago. The results are interesting, though messy. I didn’t do any advanced graph calculations so the following is all just based on visual inspection.

Some basic insights:

  • Between the 67 nodes, we had over 750 edges
    • 1 node = individual person, 1 edge = relationship between two nodes.
  • The most isolated nodes had 0 links/edges (i.e. didn’t know anybody at the university prior to attending)

  • The most connected node had more than 30 links.
    • These were students that attended the university’s summer camp program.
  • That’s an average of 11 edges per node (i.e. the mean number of connections per person was 11).

  • 19% of the nodes aren’t connected to the main group (i.e. no nth degree connections to the largest group in the graph).

  • In general, the individuals who had the least connections, still have fewer connections. While this seems obvious, I feel like it’s still worth noting that it’s (generally) easier to add linkages than to break them, so those with more initial connections maintain their more-connected status.

  • By far the most connected group was the Tajik GBAO students. This is not surprising as the region seems to be more tight-knit, geographically concentrated and isolated, and religiously/ethnically homogenous than the other places.

  • Kyrgyz, non-GBAO Tajiks had average connectedness (due to larger geographic spread).

  • Pakistani, Kazakh, and Afghan students had low connectedness (even larger geographic spread).

This graph is interesting because I highly doubt there are many other universities that have this level of interconnectedness prior to admissions. At Yale, I knew four people prior to attending and that seemed pretty normal; a lot of my classmates knew zero to two people? Of course there were people who went to more famous high schools and probably knew 30+ people but that wasn’t very common at all.

In conclusion, I hadn’t realized how generally connected the students were but it does seem to explain some of the interesting social dynamics, post hoc.