I decided to do the analysis anyway, and one of the metrics that I pulled out was diaper brand sales, in units, by state.
When I got home and told my wife, she suggested that it was just because our toddler is in the top 98% percentile for height, and the doctor probably though she was over 3yo, and didn’t really look at the chart. I could see what the average number of months that new mom’s buy diapers for. I thought, hmm, is she right? We have diaper sales data at work. When trying to come up with a hackathon project at work recently, I reflected on the previous week when I brought my 2.5yo daughter to the ER and the doctor outright told me that she was too old to still be in diapers. I wanted to display diaper units sold by brand by U.S. Photographs from a recent assignment documenting the agricultural side of cotton.I used NodeBox to do a visualization for a recent hackathon project. Looking at two studies about value and the nature of free. (My birthday is in April and I'll gladly accept Facebook stock.) Incidentally, Facebook has announced that it may have an IPO in April of this year. The amount of information and the power to harness it is truly unprecedented, but it is easy to ignore until you see it laid out in front of you. And this is merely a social graph of me and my associations one degree away imagine the graph of all 500 million-plus Facebook users all of whom have self-generated the edges of their personal graphs and are actively adding more metadata. Add to this all the metadata such as 'likes', status updates, clicks, photos, and GPS check-ins and it becomes unnerving that so much well-organized data resides with one corporation. With almost no other data you can tell a lot about me simply from the topology of this graph.
The one on the lower left consists of people who I know and who know each other through various social networks like Twitter, while the one toward the upper right are people from the Anchorage design, advertising, and photography world. Although quite densely connected, the Anchorage group also has two noticeable clusters.
This accurately represents the social structure of the school, at least at that time. Within the younger group two sub-clusters are noticeable: one group is mostly instrumentalists and one group is mostly vocalists. Looking at the college group, older students are more sparsely connected toward the left and younger students who I met when I was working at the university are tightly connected on the right. The major groups are labeled, but within these groups subgroups are apparent. This tendency of a graph to form groups of highly connected nodes is called assortative mixing and it's something that makes social network graphs special and useful for applications from counter terrorism to politics to advertising-for instance: Graph Theory, Social Networks and Counter Terrorism (1MB PDF). The graph self-organizes into tidy chronological clusters with a handful of unconnected satellites representing groups of people who I've casually met along the way. The node that looks like it occupies a central spot is appropriately Jessica, my wife, who seems to know a lot more of my friends than I realized. I removed my own node from the graph because it's redundant-I'm by definition connected to every node. The nodes are assigned a force to repel each other, but the edges create an attraction between the nodes they connect causing a clustering effect so that nodes that are tightly connected are pulled into groups. The algorithm that decides where to place each node is based on the connections. The lines, or edges, of the graph represent friendships. Each circle, or node, in the graph represents a person. I removed the names to protect the innocent (and paranoid), but left the connections between people. It was created using nodebox, one of my favorite utilities for creating visualizations. The above image is a graph of my Facebook friends and the friendships between them.
0 kommentar(er)
