Last Saturday (6/5), I participated in my first ever hackathon. It was National Day of Civic Hacking, an event that brings together civic activists, government staff, developers, and designers to “build new solutions using publicly-released data, technology, and design processes to improve our communities and the governments that serve them.”
I took part in the DC organized event, which focused on helping small businesses in the area. My team was tasked with helping local food truck businesses (an issue near and dear to my heart and stomach).
There are way more food trucks than designated parking spots in DC; this leads to a lot of inefficient roaming around and trying to figure out where to go. Our goal was to use hyperlocal, real-time(ish) data to help these roving businesses figure out the best locations to go throughout the day.
We ended up building 2 things to address this (in very half-baked forms). The first was an aggregator of big, local events in the area, laid out to give food truck owners a simple itinerary for the day (delivered via an email each morning). The second thing was a map of DC that showed people who were tweeting about various food related topics in the area (for example, where exactly are the people who are tweeting about being “hungry” or “grabbing lunch”).
I thought this second piece was really interesting but we didn’t have time to fully flesh it out. All we did was plot points (that represented individual tweets) on a map. From a user perspective, it was a bit overwhelming to consume and too difficult to draw conclusions about tweet densities in an area (because of overlapping points).
After the hackathon ended, I put in a few more hours to do the things I didn’t have time for during the event, namely cleaning up the design and showing tweet activity stats by neighborhood. I used this dataset to get the DC neighborhood geo boundaries and this point in polygon Leaflet API to assign tweets with a latitude and longitude to their enclosed neighborhood.
The final product is here (unfortunately, it seems most tweets aren’t geotagged hence the lower than expected totals). It’s still quite basic, but you can search for any word or phrase, zoom in to a particular neighborhood (either by clicking on the map or list), and click on a point to read the actual tweet. Give it a whirl!