-
-
Screenshot of code running in repl.it
Inspiration
We all knew family and friends who were not taking COVID-19 seriously and were not taking adequate safety measures, insisting that they wouldn't make a difference and that the whole thing would blow over eventually. We’ve found some simple simulations online, but most of them are not very detailed and don't allow the user to interact with the program. On top of that, we have a custom-made machine learning model to predict future infection rates. We’ve adapted it to more accurately predict an epidemic than classic machine learning models. We decided to make a simulation that could convince people through a simple User Interface and visual representation of data that preventative measures pay off against disease spread.
What it does
Simulates disease spread among a system of communities and how precautionary measures can make a huge difference, as well as provides information in the form of a website.
How I built it
My team and I collaborated in repl.it to write the main bulk of the java program; we then split into two groups: one refining the java simulation and one creating the website around it.
Challenges I ran into
There were a plethora of times when positioning of buttons and sliders did not go as planned due to the panning feature changing the relative origin of the canvas.
Accomplishments that I'm proud of
The simulation of particle systems was particularly hard to implement, and the general formatting of the page took quite some time to accomplish. The machine learning algorithms predicting the graph trends were our greatest feat in terms of results, difficulty, and new learning required to accomplish it.
What I learned
How to derive machine learning algorithms related to an epidemic, how relative canvas origin works in processing, and how to make a system of particles interact using physics and vectors. We all also learned a lot about class inheritance and nested classes while implementing the particles class as part of the communities class.
What's next for Simulation-Charter-Hacks-
We are hoping to add greater customization complexity and options, as well as to polish off the User Interface to make it more seamless and presentable. We were also thinking of refining the machine learning algorithms to increase their accuracy, as well as adding more variables to be calculated such as herd immunity. The ability for the user to zoom in and out of the web page and simulation itself is another thing that we plant to add. Eventually, we were going to port this to an app as well so that we can gain a user base and improve it based off of feedback.
Log in or sign up for Devpost to join the conversation.