This project aims to provide detailed insights into geographical trends that aren't easily available elsewhere. For example, most political maps are flat, making it easy to misinterperet. These are in 3D, so you can easily tell how many people voted in an area. While shootings don't encompass all of violent crime, they can be viewed as a proxy for it. Demonstrating shootings in this fashion allows the user to quickly determine where the hot spots are, as well as understand its spread.
You can pan around, zoom in and out, and some layers are interactive. Hover over a recent shooting and a popup will show information about the incident. Click or touch a hexagon on the gun violence map and a popup will show the total shootings, and a graph by year. The political map users can click or touch a hexagon to generate a popup that shows the total vote count, and which 2020 candidate won and by how much. Zoom in to see a random distribution where one vote = one colored dot. A hover here shows the precinct level breakdown.
The data sources are: Gun Violence Archive, Overture Maps Foundation, The Upshot, openaddress
We use various common spatial analysis techniques including clustering, interpolation, normalization, and network analysis to derive meaningful patterns from the data.
We'd like to. Modern AI techniques are making it more possible to glean incredible insights from data that's surrounding us every day. If it's insightful and there's a use case, we'll try to incorporate it.
We use common coloblind accessible gradients for both the hexagons and buildings/dots. The heights on the hexagon levels for the gun violence map show the overall risk. The shootings have been normalized by population. The colors also correspond to risk. For Political Leanings, the height of the hexagons correspond to total votes in the area, and the colors correspond to which party won, and by how much. Detailed maps show a single dot for each vote cast in the 2020 general election, randomly placed inside the voting precinct. This dot density method allows a user to approximate the number of votes as well as their distribution.
While there may be some missing data in both Gun Violence and Political Leanings sources (e.g., some states don't report their voting precincts), the methods for assigning risk scores to well vetted, and are unlikely to contain an error unless the sources are incorrect or missing. Realbloc updates the source data regularly