covid-19-timelapse

COVID-19 Soluttion Pattern

As the COVID-19 pandemic evolves, industry leaders require a holistic view of the issue to be able to make smart decisions.

This pattern leverages Dow Jones and public datasets to build a time-lapse dashboard that shows COVID-19’s global impact on people, industries and regions. It uses data visualization tools and data science techniques to identify industry news sentiment and coverage patterns to achieve this.

Solution

The solution consumes Dow Jones news and industry datasets via the Factiva Snapshots API.

The data for this solution comes from developers’ first-party industry lists and the following third-party public dataset:

Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU)

This solution is built inside a Heroku container, which hosts a Flask App for data processing and a Dash app for display. Graph objects are being synced/connected via a single controller within the Dash app and mapping tiles are provided via Mapbox.

Bi-gram frequency is calculated using Natural Language ToolKit (NLTK) and sentiment is calculated using a simple Text-Blob approach in order to reduce processing time.

Architecture

The application back-end processing is done by leveraging Python to extract, clean, enrich and aggregate files. This processing enables clean input for the solution’s dashboard.

Installation

• Install python 3 and pip

• Install the dependencies from requirements.txt

• Run pip install -r requirements.txt

Run

• It is recommended to have a virtual environment
• From cli run:

    virtualenv venv 
    source venv/bin/activate
    
    gunicorn app:server