Authors

Inspiration

We are looking at probably one of the most confusing and devastating time periods in recent history. Many people are dying and we wanted to do what can to help so we decided to make a public domain for the government and really anyone that wants to know more about the virus so they can possibly prepare for it.

What it does

We are providing Coronavirus data analysis and forecasting while using machine learning models to predict future cases.

How we built it

Our website was created using bootstrap and firebase along with basic HTML, CSS, and Javascript. For the ML/Data Science Portion, there are three models that we deployed: RNN, Facebook Prophet, and Exponential Regression.

Challenges we ran into

We were trying to do more with the predictions using firebase and firestore but lack of experience with databases and server side programming made it difficult to finish. In addition, it was our first time using RNN and Facebook's Prophet Library so we had to get accustomed to optimizing the parameters and using the specific metrics within Prophet.

Accomplishments that we're proud of

We don't really have any experience with web editing and we made basically almost all of the website from scratch apart from tiny uses of a bootstrap template. It took hours of coding which we feel like we learned a lot about. For ML, we have never used a recurrent neural net or prophet so understanding the theory and applying it in the span of two days was something we did not think we can do.

What we learned

We learned how tedious creating websites could be and how various machine learning models worked.

What's next for Coronavirus Forecasting Dashboard

We hope to be able to make this into a live dashboard where we can take into consideration other factors such as temperature to make it even more accurate so that the government and other officials can put the results to use in regards to allocating an appropriate amount of resources.

Try It out

Hackathons

Technologies

basemap, bootstrap, css3, firebase, html5, javascript, keras, machine-learning, plotly, python, sci-kit, tensorflow

Devpost Software Identifier

256680