Covisualize

A full-stack website to track real-time COVID-19 cases and deaths and predict them for the next one month using time-series analysis and SARIMAX model.

Working Demo

Data Collection

The data for COVID-19 was obtained using disease.sh - Open Disease Data API For example-

On hitting the endpoint- https://disease.sh/v3/covid-19/all

Response-
{
  "updated": 0,
  "cases": 0,
  "todayCases": 0,
  "deaths": 0,
  "recovered": 0,
  "todayRecovered": 0,
  "active": 0,
  "critical": 0,
  "casesPerOneMillion": 0,
  "deathsPerOneMillion": 0,
  "tests": 0,
  "testsPerOneMillion": 0,
  "population": 0,
  "oneCasePerPeople": 0,
  "oneDeathPerPeople": 0,
  "oneTestPerPeople": 0,
  "activePerOneMillion": 0,
  "recoveredPerOneMillion": 0,
  "criticalPerOneMillion": 0,
  "affectedCountries": 0
}

The deaths and cases are obtained from here for the past 1 year.

Database and APIs

DynamoDB has been used as a NoSQL database for storage and retrieval of the data fetched from the disease.sh API. Then AWS Lambda has been used for making API endpoint for the retrieval of predicted cases and deaths for visualization in the UI.

Machine Learning

The SARIMAX time-series model is used for predicting the cases and deaths for the next one month given the past data. AWS Sagemaker's Jupyter instance was used to train the model and store the predicted cases and deaths in the DynamoDB.

Deployment

The whole application was packaged and deployed to the web using Amazon S3

Run frontend locally

To see the UI and landing page locally, run the following commands one by one

cd client
yarn
yarn start