Search
  • Nikki

How Steamboat stays connected: Mapping Facebook’s Social Connectedness Index


It's March again. There is not much left to say about the COVID-19 pandemic that I can do a better job of saying than anyone else, so instead, I made some maps. This month, as an ode to this past year of confusion and isolation, our data topic is not anything natural resources or nature related. Because this past year, we've needed connection in any way we can find it. And social media companies like Facebook have made it kind of mandatory that we find it there.


So this month, we look at Routt County's Facebook friends around the country. Who we're most connected to, and where our social media blind spots are.


Facebook's Social Connectedness Index (SCI) in relationship to Routt County.


We keep our friends close

Honestly, this map surprised me at first glance. Steamboat is a pretty transient place, or at least, so I thought. I assumed our connections would be more evenly spread around the country. Instead, one of those universal truths of geography seems to have its hold on this little ski community: we have the most connections to people closest to us in space. I guess humans are spatially autocorrelated too!


Here's another way of breaking the color scheme to bring out those hot and cold spots a little more.

The data are highly skewed with a large range. Here's what Facebook says about what those huge numbers actually mean:

"the 𝑆𝑜𝑐𝑖𝑎𝑙 𝐶𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑𝑛𝑒𝑠𝑠 𝐼𝑛𝑑𝑒𝑥 i,j measures the relative probability of a Facebook friendship link between a given Facebook user in location i and a given user in location j. Put differently, if this measure is twice as large, a given Facebook user in location i is about twice as likely to be connected with a given Facebook user in location j. We also add a small amount of random noise and round the Social Connectedness Index to the nearest integer, to ensure that no single individual or friendship link can be identified from the data.


Our faraway friends are from some familiar places

Holy Vermont and New Hampshire! And what's up Lake Tahoe mountains. And Upper Midwest, we know you love you some Colorado. It's not really a surprise that some of the other "hot spots" are in places where people tend to migrate to Steamboat from, either temporarily or "I came for one winter in '86 and never left" temporarily. These patterns are comforting to me. They remind me of all the chairlift rides with strangers and the 4-minute life stories they leave you with.


Still can't explain the South Florida connection though. I love it: someone tell me why you're hoarding Facebook friends in Miami.

Learn data science fundamentals by

following along with this post

DATASET: Facebook's Social Connectedness Index

LANGUAGE: R

LEARN: Choropleth maps, Mapdeck widgets

Tutorial

But this isn't the ideal way to visualize network data. We can have way more fun with it. And I DID. I made an interactive Mapbox widget of the top 10 percent of Routt's SCI connections.




From my brown and green forest in the Southeast, these little arcs makes me feel that much closer to your white and green forests, Steamboat. And hey, Dare County is in the top 10 percent of Routt County's Facebook friendship connections, so when I head to the NC coast, I'll know I'm with my people.


Friends forever.


The Details

I will not sing the praises or decry the evil of social media in this post. I'll stay Facebook-neutral, here. But I am always pro-open data, and am always encouraged to see private companies sharing what data they can and trying to do good with it. Do good in the world, perhaps, but also give me more spatial data to play with.


This dataset is honestly very clean and straight-forward. And massive. Maybe not by "big data" standards, but 10,426,441 entries (one value for each county-county connection) is a healthy dataset.


It's size and tidiness make it a wonderful starting point for learning how to make those interactive Mapbox maps like I did above. It's completely free, and I've made it cut-and-paste easy (hopefully). So if you want to recreate these maps, dive into the data on your own, or just see what it takes to create these, head over to the tutorial on Github. There is also a really fun animated version of the arc map over there that looks like an erupting fountain and serves no other purpose whatsoever.



Check back for the next post on April 6.

0 comments