Simulation study of various variable selection techniques. Intended to examine the bias in parameter estimation in the context of causal health. In particular, we grapple with confounding, bias amplification, missingness and other features important to causal inference.
The folders "Jupyter", "data" and "plots" contain pre-computed results.
The simulation can currently be run in one of two ways:
- In R / RStudio using the script "Run_Sim.R"
- In Jupyter Notebook via the R Kernel using the notebook "Run_Sim.ipynb"
Still in development!