Business challenge : CoronaRisk

A data-driven roadmap to assessing covid risk in every Dutch region : Through multiple linear regression, we provide travellers with an overview of risk areas by postcode

Comment

Inspiration

  • Coronavirus is here to stay
  • The Netherlands is entering a fourth wave of infections
  • Due to summertime and vaccinations, measures have been widely lifted
  • The Delta variant seems to be the dominant variable in the Netherlands and not sufficient data are still there to support that vaccinations provide full protection against it
  • Therefore,it is imperative that people know the risk when travelling to a specific region of the country

What it does

This app runs on a model that predicts a risk level from "On track for containment" to "Tipping point" based on the data of a region. The user enters a postcode and the risk level and corona stats from the region are displayed. While the "risk level" is normally measured through the number of cases / population ratio, we thought there are many more factors to be considered. We propose training a model on all the stats concerning the pandemic in a specific region and assigning our own risk scale according to the current data.

How we built it

For this mock-up we generated our own dataset: this is a placeholder for the real data which would be eventually pulled from RIVM. We wanted something to prototype and test our idea quickly and create a solid proof of concept, and we also wanted to find out how we can generate (partially randomised) data, since this practice can have many speculative applications.

Challenges we ran into

Finding a valid postcode dataset was a major challenge: all APIs we found are paid and the only CSV dataset we could use for free has proven to be incomplete. For this reason we suggest to test the application with postcodes from the CSV file.

What's next for Business challenge : CoronaRisk

Next step would be shifting our application from this realm of speculative data to a real-time, reliable dataset from RIVM. A model such as this one would need to be re-trained periodically since new variants or number of vaccinated people might change the way the risk level relates to the stats continuously

Built With

Share this project:

Updates