nopanic.io

Inspiration

Amidst the ongoing COVID19 outbreak, we felt a tool was needed to provide quality and fact-based - but also contextual - information in order to help mitigate the fear factor associated with such events. We wanted to offer this solution in a dashboard-type web application, all the while making it visually appealing and staying far from the bright reds and blacks, which are usually more fear-inducing than calming.

What it does

Nopanic.io offers many features. First of all, the user is able to see the different COVID19 outbreaks around the world for the present, the past, and even 7 days in the future using a machine learning model to predict the future number of cases. Although getting the big picture is always useful, we wanted to provide the user with a way to know if they have to adapt their lifestyle depending on their region's situation. Thus, by entering their location in the search bar, the user is able to see the number of cases in their region and, based on this number, is given safety guidelines and useful hotlines. For example, if the number of cases in an area is under 100, the user is told to proceed normally but to adopt good sanitary discipline. However, if the number of cases exceeds 1,000, the user better avoid using public transit and stay at home until the outbreak calms down.

How we built it

We extracted our data from the John Hopkins University public COVID19 open-source dataset made available on Github using pandas. We then transformed this dataset into daily sub-sections and exported them into .geojson files in order to be able to visualize them on Google Maps. This transformed data was then uploaded to a Google Cloud Storage Bucket. At the same time, we also built a machine learning model using linear regression, predicting the potential number of infected victims within the next week across all regions of the world. The frontend was built using the Flask python framework, which loaded a map after sending a request to the Google Maps API. The styling and utilities that are on the screen have all been made using scratch CSS, HTML and Javascript. Finally, Matplotlib was used to visualize the increase in the number of people infected across the years.

Challenges we ran into

We ran into a few challenges when building nopanic.io. The biggest obstacle was definitely writing the Javascript and working with the Google Maps API, since we both had no experience with it beforehand. Another challenge was writing the Machine Learning Model since both of us had more experience with designing especially deep learning models, instead of simpler models that use Linear Regression for example.

Accomplishments that we're proud of

We are very proud to have a robust machine learning model which allows us to offer some very valuable insights. Furthermore, the backend heavy lifting has not prevented us to design a beautiful user interface, which is also a great source of pride.

What we learned

Lucas: This was my first time building a full-fledged web application, and I had not coded in more than six months. Although a bit rusty at first, I was able to get into a rhythm building the front-end, as well as the Flask integration. I ran into a few challenges along the way, but it allowed me to see how coding projects are really undertaken. Lastly, I discovered the power of dividing and conquering while working on such projects: not only were we able to create a better solution but it incorporates both of our strong suits like design and data manipulation.

Steven: This was also my first time building an entire web application to production, working with online apps and sending API requests using custom keys. I really enjoyed working with Lucas who has shown me his great talent in creating incredibly beautiful frontend designs.

What's next for nopanic.io

We have a few more features in the pipeline for nopanic.io. For example, we would like the resources and safety guidelines to be in sync with those offered by local authorities. We also envision adding some more graphing and prediction tools in order for users to monitor the outbreak while staying informed of the steps to follow.