COVID-19 Machine Learning Based Calculated with Map

Global Distribution of COVID-19 Cases

Inspiration

We were inspired by the COVID-19 pandemic and wanted to develop a solution for people to understand how different factors affect an individuals chances of getting the COVID virus.

What it does

We created a machine learning based calculator that predicts the probability of contracting COVID-19 based off of your county and state case history.

How we built it

We began by developing a python script to graph the COVID cases on around the world using a dataset from Johns Hopkins University. We were able to refactor the data and use it to find the likelihood of getting COVID based on your state and county and then compared that to the national average to determine a low, moderate, or high probability of getting COVID. National averages were determined using a machine learning Bayesian network analysis in PyMC3 on county population and cases in the last 7 days. Our model was fit with a linear regression and our training model was validated by analyzing the validity and accuracy of each parameter. Overall we believe we have created an accurate prediction of contracting the COVID 19 virus based off of your county and state.

Challenges we ran into

Some difficulties we ran into was using Machine learning to find the trends in covid cases as it varies with different populations. My team was very new to machine learning and had trouble finding documentation that was beginner friendly. Through a lot of trial and error we were able to fine the slope of the trend line and calculate the national average of covid cases based on population. We also had trouble finding a suitable collaboration resource that allowed everyone to upload their code and run code without errors, we ended up working in google Collab for a majority of the work so we were all on the same page in terms of environment.

Accomplishments that we're proud of

This was my team and I's first experience with machine learning and we were very proud to have been able to successfully program this calculator.

What we learned

We learned about Machine learning, Bayesian networks, and how to use various plotting packages for data science and machine learning applications.

What's next for COVID-19 Machine Learning Based Calculated with Map

In the future, we hope to link our calculator to an Android or IOS app and have it continuously update to provide the latest data related to COVID-19. We hope to integrate other functionalities and outfit a multivariate linear regression model to make predictions based off of additional parameters such as race, age, sex, socio-economic status, among many other things.