/pycovid

Python workspace for COVID-19 analysis

Primary LanguageJupyter NotebookGNU General Public License v3.0GPL-3.0

pycovid

Author(s): Luke Boegner

Project Overview

Applying data analysis and machine learning to better understand various aspects of COVID-19. Primary current objective is to better understand how US State government actions have been affecting the spreading rate of the virus. Current approach is to read text from Wikipedia to extract information for each state's responses, build a machine-learning (ML) natural language processing (NLP) model to interpret the severity of various actions, and then plot the actions against the growth rate of the confirmed cases. Follow-on analysis can then be done to determine if/when different levels of actions helped slow down the spread of COVID-19.

TODO:

  • Continue manually labelling wikipedia data from each state's COVID-19 pages ( running label_wiki_data notebook and manually labelling sentences)
  • Enhance NLP model once more data is generated above (running model_dev notebook)
  • Integrate NLP ML model with plotting of data (currently hardcoded in covid_analysis notebook)
  • Perform follow-on analysis of the results seen from the plotting
  • Transition to using wikipedia's confirmed cases data instead of the JHU CSV