Inspired to do something related to data visualization, we turned to construction of Voronoi Diagrams, famous for their applications in numerous academic and industrial fields. In this case, Capital One's Hackathon API provided valuable data for us to use.
What it does
This 'hack' allows a company to visualize which vendors customers are using most through the use of voronoi diagrams. Currently, this hacks draws data about users and merchants from the CapitalOne Hackathon API and uses the HERE Geocoder API to convert addresses into lat/long coordinates. In this version in our explorations of the Voronoi, we mapped Vendors to numerous CapitalOne accounts, and showed how certain vendors are not used at all in our subset of data, and how some vendors seem overcrowded, possibly needing more attention by the company.
How we built it
Nikhil used Java to construct the Voronoi Diagrams. Each '$' is a vendor, and the points of the same color are within its voronoi cell. Kazu learned how to use the CapitalOne Hackathon API and HERE's Geocoder API to effectively gather, filter, and produce data to be processed by Nikhil. He used Python to gather and parse through the data.
Challenges we ran into
Constructing voronoi diagrams is very difficult with no test set of points. Although we originally split up the workload to improve efficiency, occasionally we were unable to progress until we waited for the other to finish their part. Other than that, mathematically, voronoi diagrams are difficult to draw correctly. the construction of the edges in itself took most of our hackathon time. Similarly, learning new APIs is always a challenging and time-consuming task as one must read the documentation carefully to avoid errors.
Accomplishments that we're proud of
Our voronoi diagram honestly looks really cool.
What we learned
We learned a lot about how cool visualized data is. Aside from that, planning was probably a big learning step for us. We were not experienced enough to 'hack' together a project as Freshmen. This is mainly due to the scope of our project and the learning curve involved with team projects. But in the end, we learned how to create cool looking voronoi diagrams, gather data from online APIs, and have lots of fun with in the lair of DDOSKI!
What's next for Data Density Visualizer
With some more time, we hope to apply our hack to other uses, such as visualizing coffee shops and coffee drinkers to make supply shipments more efficient (since our data visualization would show which shops are most visited). Furthermore, we could consider more factors instead of distance in calculating other attributes of a certain shop. For example, we could consider hiring more employees at a location with a lot of traffic compared to other locations.