Setup for a Causal Analysis of Covid19 Data

less than 1 minute read


In this post we present a mock setup for performing causal analyses on covid19 data using the dowhy library for causal analysis.

We focus on the question How is the implementation of existing strategies affecting the rates of COVID-19 infection?, which has a clear causal nature. We illustrate the steps of data extraction, model definition, identification of causal estimands, estimation and refutation.

The data we consider is very limited, and the model extremely simplified and not realistic. Proper modelling would require knowledge in the domain of epidemiology and sociology. Thus the conclusion reached at the end of this notebook have NO real-world relevance. The main point of this notebook is to illustrate a preliminary setup for causal analysis.

Notebook here