Inspiration
Gentrification is a huge problem that often causes prices to rise in many neighborhoods and often leads to renters being priced out of the residential district. This web app aims to use demographic data to predict the areas of a city using census tracts that are the most at risk of being gentrified. The name "ToastBusters", is a testament to the fact that Avocado Toast is often a symbol of gentrification of an area, and ToastBusters in based on the idea of trying to better control gentrification, like GhostBusters.
What it does
This app currently displays the census tracts of Atlanta and uses demographic data to predict the vulnerability of a certain area of being gentrified. This is used by taking in several sets of demographic data based on 2010 census tracts and feeding that information through a neural network in order to predict the vulnerability of an area on a scale of 0-4(0 means no vulnerability, 4 means high vulnerability).
How I built it
The neural network training model was created in Python and using the SKLearn and Python libraries. The frontend was hosted on a Flask server and was developed using Bootstrap and the JavaScript SDK API of Google Maps.
Challenges I ran into
The biggest challenge that we ran into was displaying the census tracts as a polygon in the API of Google Maps. In addition, data was often hard to obtain and trying to display census tracts as a polygon on the map was difficult moslty since the data was not in the proper format(json) or not available period.
Accomplishments that I'm proud of
We are proud of the model that we built and the predictions it was able to make. Our model with training data was around 85% accurate, which is fantastic for a neural network especially given the small sample size of the data that was used.
What I learned
We learned how to use the Google Maps API and display polygon overlays on the overlay. We also learned how to create a neural network and use hidden layers to be able to use data to predict values.
What's next for ToastBusters
Expansion to other cities in the United States that tend to have huge gentrification problems such as San Francisco as well as consistently improving models in order to refine the predictions.
Log in or sign up for Devpost to join the conversation.