Encoding Causal Macrovariables

Data

• Natural climate data ('El Nino')
• Self-generated data ('Simulated')

Experiments

1. Detecting macrovariables through the Causal Autoencoder (in the code termed VCE)
2. Transforming detected variables through double VAE and analysing causal relationships through additive noise model (ANM)

Repo Structure

• Four python scripts to run experiments
• Two jupyter notebooks to visualise results
• src containing code base, helper functions and modules
• data with two subfolders for the two datasets
• results with two subfolders for the experiments on the two datasets