SEE:Net - Network graph of MIT's Social Evolution Experiment
SEE:Net is a window into the social lives of a group of MIT Freshmen.
Alex Pentland's lab at MIT used phone and sensor data to closely monitor the physical and social movement of a group of Freshmen, for a whole year. They made their datasets public, and my team and I thought it would be a great opportunity to showcase the different ways that network graphs can be visualized. (Teammates: Brian Feeny, Jennifer Sulkow)
Data visualization is often underestimated as a tool for discovery in data science. Visuals are frequently an afterthought to the "real" work of modeling and prediction, things to put in a slide deck to show the management. That's a mistake, in my opinion - especially when the data you're dealing with are complicated and hard to think about.
Network graphs are a great example of data that human brains just don't represent very well, and there's real value in the exploratory phase of a project in finding good ways to visualize them.
SEE:Net shows three different views of the SEE data, using D3:
We also added a simple animation function, so you can watch how the relationships in the network evolve over time. Combining multiple views with dynamic time series information yields a visual representation of the data that would be very hard to come up with otherwise.
Check out our screencast for a brief introduction to SEE:Net.