At one point or another, we've all wondered where Colorado Springs music listeners fit into the hierarchical clustering, geographical flow, and time-lagged correlation of Euclidean distances between cities in a normalized listening matrix.
According to a new University College Dublin study entitled "The Geographical Flow of Music," whatever I wrote in the previous sentence can be ascertained by adopting "a method previously used to detect the leadership networks present in ﬂocks of birds."
What the study does — I think — is to use last.fm statistics to determine which cities are the biggest music trendsetters. As you may have guessed, Colorado Springs never quite finds its way into the matrix.
On the other hand, as you can see in the charts above, Denver has clawed its way up to the bottom-rung in the hip hop, indy, and general music categories.
The study also goes on to posit a scenario in which there are only two artists, Radiohead and Coldplay, and two cities, Los Angeles and Seattle — which I believe is being considered as a sequel to Cormac McCarthy's "The Road."
Anyway, you can read the whole study here, or just stare blankly at the following excerpt and then go do something else.
To construct the dendrogram shown in ﬁg. 1, we performed average linkage clustering (an agglomerative clustering algorithm) on a distance matrix D of the cities, a square matrix where each entry Di;j is the Euclidean distance between city i and j. Instead of constructing the dendrogram based on just a single listen matrix, we summed together the distance matrices associated
with the all of the listen matrices in our dataset. The colored clusters are the result of taking a ﬂat cut to the dendrogram at a height which we chose manually.