Correlations between Massachusetts Cities and Towns
A Tribute to Google Correlate
As a tribute to Google Correlate that shut down on December 15th, 2019, “as a result of low usage” (sic!), I prepared a map of all cities and towns of the Commonwealth of Massachusetts.
I started by extracting the most significant correlations between the queries referring to each city or town (there is no data on the map for the “pale” towns in Western Mass). For example, Lexington-related queries are strongly correlated with Arlington-related queries, presumably because the two towns are neighbors and have similar demographics. I built a network of cities and towns, using pairs of high correlations as edges.
I extracted six dense communities (clusters) of towns and cities using the Louvain community detection algorithm. The clusters are elongated in the longitudinal direction, roughly corresponding to the North Shore, South Shore, Southeast, and Berkshires.
Surprisingly, Cape Cod and the Islands at least partially belong to the Berkshires clusters — possibly because all of them are perceived as recreational destinations. We will never know for sure. R.I.P., Google Correlate.
Read more about complex networks, their construction, and analysis in my book Complex Network Analysis in Python. Recognize → Construct → Visualize → Analyze → Interpret.